Ovarian cancer: the recognition and initial management of ovarian cancer

Transcription

1 Clinical Guideline Ovarian cancer: the recognition and initial management of ovarian cancer 0 Full Guideline The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

2 Contents Foreword Key priorities Key research recommendations Methodology Algorithm Epidemiology 0. Introduction 0. Data collection 0. Incidence. Mortality. Survival. Routes to diagnosis. Treatment. The findings of cancer peer review of gynaecology cancer teams in England Summary Detection in primary care. Awareness of symptoms and signs. Asking the right question - first tests Establishing the diagnosis in secondary care. Tumour markers: which to use?. Cancer pathway management: malignancy indices. Imaging in the diagnostic pathway: which procedure and where?. Tissue diagnosis Management of suspected early stage ovarian cancer. Staging: The role of systematic retroperitoneal lymphadenectomy. Adjuvant systemic chemotherapy in stage I disease: patient selection Management of advanced stage (II-IV) ovarian cancer. The value of primary surgery. Intra-peritoneal chemotherapy. Chemotherapy regimens Support needs for women with newly diagnosed ovarian cancer Appendix : A cost-utility analysis of diagnostic investigations in primary care for women with symptoms of ovarian cancer Appendix : Abbreviations 0 Appendix : Glossary Appendix : Guideline scope 0 Appendix : List of topics covered by each chapter Appendix : People and organisations involved in production of the guideline The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

3 0 0 Foreword These clinical guidelines review a number of clinical questions that involve the detection, diagnosis and initial management of primary epithelial ovarian cancer and which focus on areas of uncertainty or where there is a wide variation in clinical practice. The clinical questions were chosen using a consultative process that involved an array of stakeholders that included patient groups, representatives from relevant professional organisations and the pharmaceutical industry. For each chapter of the guideline, the Guideline Development Group (GDG) have made evidence-based recommendations concerning clinical practice and, where applicable, some recommendations on future research. The GDG are pleased that the focus of many of the clinical issues relate to an early stage in the patient pathway with particular relevance to patients and their families. In particular, identifying the first tests in primary care should help ensure women are directed onto the right clinical pathway in a timely fashion. The chair and lead clinician were aided and supported by a diverse and engaged GDG membership whose complementary skills and perspectives have been instilled in this guideline. Mr Sean Duffy, GDG Chair Mr Charles Redman, GDG Lead clinician The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

4 0 0 0 Key priorities Awareness of symptoms and signs Carry out tests in primary care (see section. on page ) if a woman (especially if 0 or over) reports having any of the following symptoms on a persistent or frequent basis particularly more than times per month: o persistent abdominal distension (women often refer to this as bloating ) o difficulty eating and/or feeling full (early satiety) o pelvic or abdominal pain o increased urinary urgency and/or frequency. Carry out appropriate assessments for ovarian cancer (see section.. on page ) in any woman of 0 or over who has symptoms that suggest irritable bowel syndrome (IBS) because IBS rarely presents for the first time in women of this age. Asking the right question - first tests Measure serum CA in primary care in women with symptoms that suggest ovarian cancer (see section. on page ). If serum CA is greater than IU/ml, arrange an ultrasound scan of the abdomen and pelvis. Advise any woman who has normal serum CA, or CA greater than IU/ml but a normal ultrasound, to return to her GP for re-assessment if her symptoms persist. Malignancy indices Calculate a risk of malignancy index I (RMI I) score (after performing ultrasound; see section.. on page ) and refer all women with an RMI I score of 00 or greater to a specialist multidisciplinary team. Tissue diagnosis Obtain a confirmed tissue diagnosis before offering cytotoxic chemotherapy to women with suspected advanced ovarian cancer in all but exceptional circumstances (see section. on page ). Staging: the role of systematic retroperitoneal lymphadenectomy Do not include systematic retroperitoneal lymphadenectomy as part of the standard surgical treatment of suspected ovarian cancer in women whose disease appears to be confined to the ovaries (that is, who appear to have stage I disease). Adjuvant systemic chemotherapy in stage I disease: patient selection Do not offer adjuvant chemotherapy to women who have had optimal surgical staging and have low-risk stage I disease (grade or, stage Ia or Ib). Support needs for women with newly diagnosed ovarian cancer See Irritable bowel syndrome in adults (NICE clinical guideline ). Optimal surgical staging constitutes midline laparotomy to allow thorough assessment of the abdomen and pelvis; a total abdominal hysterectomy, bilateral salpingo-oophorectomy and infracolic omentectomy; biopsies of any peritoneal deposits; random biopsies of the pelvic and abdominal peritoneum and retroperitoneal lymph node assessment [Winter Roach et al. (00) Adjuvant (post-surgery) chemotherapy for early stage epithelial ovarian cancer. Cochrane Database of Systematic Reviews 00, Issue : CD000}] The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

5 Offer all women with newly diagnosed ovarian cancer information about their disease, including psychosocial and psychosexual issues that: o is available at the time they want it o includes the amount of detail that they want and are able to deal with o is in a suitable format, including written information if possible. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

6 Key research recommendations. Further research should be undertaken on the relationship between the duration and frequency of symptoms in women with ovarian cancer before diagnosis, the stage of disease at diagnosis and subsequent survival. Most women presenting with ovarian cancer have advanced disease and have had symptoms for months. Greater awareness among both women and healthcare professionals might result in women presenting earlier with less advanced disease, leading to better outcomes. There is insufficient understanding of the factors that influence earlier diagnosis in women with ovarian cancer, especially the relationship between duration of symptoms and stage at diagnosis. Data demonstrating benefits from earlier presentation will justify investment in raising awareness among women and healthcare professionals. This is likely to be a population-based study that records both the duration and frequency of symptoms.. Further research should be undertaken to determine the optimum RMI I threshold that should be applied in secondary care to guide the management of women with suspected ovarian cancer. Variation exists in the current evidence base as to the optimum RMI I threshold that should be applied in secondary care. The cut-off levels used will have implications for both the management options considered and the number of women who will be referred for specialist treatment. Therefore it is important to establish the relative sensitivities and specificities at the different levels. The research should be a prospective observational cohort study evaluating women referred with suspected ovarian cancer. Diagnostic accuracy, sensitivity, specificity and cost effectiveness should be examined at the different RMI I thresholds.. Large multicentre case control studies should be conducted to compare the accuracy of CT versus MRI for staging and for predicting optimal cytoreduction. Currently most women with ovarian cancer will undergo a CT scan before surgery to assess the extent and resectability of disease. CT and MRI are complementary in their abilities to detect disease, but no adequate studies have been performed that compare their effectiveness in women with suspected ovarian cancer. No comparative studies have been undertaken evaluating surgical outcome. A prospective study in women undergoing primary surgery would be feasible.. A prospective randomised trial should be undertaken to evaluate the cost effectiveness and associated risks of systematic retroperitoneal lymphadenectomy in women with ovarian cancer whose disease appears to be confined to the ovaries. Systematic retroperitoneal lymphadenectomy is an untested procedure but is likely to be more accurate than lymph node sampling, with potential benefits to the woman of avoiding chemotherapy. However, increased risks are associated with it. Although there may be no overall survival advantage of this procedure, avoidance of chemotherapy and impact on quality of life may make it attractive to some women as a treatment option. In order to counsel women appropriately it is essential to understand fully the risks associated with this surgery as well as the benefits. Researchers should be encouraged to develop a prospective randomised trial with international collaboration to answer this question in a timely manner. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

7 0. Research should be undertaken to determine the effectiveness of primary surgery for women with advanced ovarian cancer whose tumour cannot be fully excised. Most women with advanced ovarian cancer undergo surgery at some point. Previous studies have shown that surgery after the completion of chemotherapy has no therapeutic value. Studies are being performed to investigate whether the timing of surgery during primary chemotherapy influences outcome. No studies have evaluated whether primary surgery itself has any therapeutic value when compared with chemotherapy alone. The potential advantages of surgery have to be offset against the undoubted costs, morbidity and occasional mortality. This would be a prospective randomised clinical trial recruiting women who have biopsy-proven advanced ovarian cancer and who are fit enough to receive surgery and chemotherapy. Women would be randomised to either chemotherapy and surgery (conventional arm) or chemotherapy alone (experimental arm). Primary outcome measures would be survival at and years. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

8 Methodology Introduction What is a Clinical Guideline? Guidelines are recommendations for the care of individuals in specific clinical conditions or circumstances from prevention and self-care through to primary and secondary care and on to more specialised services. NICE clinical guidelines are based on the best available evidence of clinical and cost effectiveness, and are produced to help healthcare professionals and patients make informed choices about appropriate healthcare. While guidelines assist the practice of healthcare professionals, they do not replace their knowledge and skills. Clinical guidelines for the NHS in England, Wales and Northern Ireland are produced as a response to a request from the Department of Health (DH). They approve topics for guideline development. Before deciding whether to refer a particular topic to the National Institute for Health and Clinical Excellence (NICE) they consult with the relevant patient bodies, professional organisations and companies. Once a topic is referred, NICE then commissions one of four National Collaborating Centres (NCCs) to produce a guideline. The Collaborating Centres are independent of government and comprise partnerships between a variety of academic institutions, health profession bodies and patient groups. The National Collaborating Centre for Cancer (NCC-C) was referred the topic of the recognition and initial management of ovarian cancer in October 00 as part of NICE s seventeenth wave work programme. However, the guideline development process began officially in February 00 when sufficient capacity became available at the NCC-C. Who is the Guideline Intended For? This guideline does not include recommendations covering every detail of the recognition and initial management of ovarian cancer. Instead this guideline has tried to focus on those areas of clinical practice (i) that are known to be controversial or uncertain; (ii) where there is identifiable practice variation; (iii) where there is a lack of high quality evidence; or (iv) where NICE guidelines are likely to have most impact. More detail on how this was achieved is presented later in the section on Developing Clinical Evidence Based Questions. This guideline is relevant to all healthcare professionals who come into contact with patients with ovarian cancer or suspected of having ovarian cancer, as well as to the patients themselves and their carers. It is also expected that the guideline will be of value to those involved in clinical governance in both primary and secondary care to help ensure that arrangements are in place to deliver appropriate care for the population covered by this guideline. The Remit of the Guideline Guideline topics selected by the DH identify the main areas to be covered by the guideline in a specific remit. The following remit for this guideline was received as part of NICE s seventeenth wave programme of work: To prepare a clinical guideline on the recognition and initial management of ovarian cancer, to include both surgery and chemotherapy. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

9 Involvement of Stakeholders Key to the development of all NICE guidance is the involvement of relevant professional and patient/carer organisations that register as stakeholders. Details of this process can be found on the NICE website or in the NICE guidelines manual (NICE 00). In brief, their contribution involves commenting on the draft scope, submitting relevant evidence and commenting on the draft version of the guideline during the end consultation period. A full list of all stakeholder organisations who registered for the recognition and initial management of ovarian cancer guideline can be found in Appendix.. The Process of Guideline Development Who Develops the Guideline? Overview The development of this guideline was based upon methods outlined in the NICE guidelines manual (NICE 00). A team of health professionals, lay representatives and technical experts known as the Guideline Development Group (GDG) (see Appendix.), with support from the NCC-C staff, undertook the development of this clinical guideline. The basic steps in the process of developing a guideline are listed and discussed below: using the remit, define the scope which sets the inclusion/exclusion critera of the guideline forming the GDG developing clinical questions developing the review protocol systematically searching for the evidence critically appraising the evidence incorporating health economic evidence distilling and synthesising the evidence and writing recommendations agreeing the recommendations structuring and writing the guideline updating the guideline. The Scope The remit was translated into a scope document by the Guideline Development Group (GDG) Chair and Lead Clinician and staff at the NCC-C in accordance with processes established by NICE (NICE 00). The purpose of the scope was to: set the boundaries of the development work and provide a clear framework to enable work to stay within the priorities agreed by NICE and the NCC-C and the remit set by the DH inform professionals and the public about the expected content of the guideline. provide an overview of the population and healthcare settings the guideline would include and exclude specify the key clinical issues that will be covered by the guideline inform the development of the clinical questions and search strategy Before the guideline development process started, the draft scope was presented and discussed at a stakeholder workshop. The list of key clinical issues were discussed and revised before the formal consultation process. Further details of the discussion at the stakeholder workshop can be found on the NICE website ( The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

10 The scope was subject to a four week stakeholder consultation in accordance with processes established by NICE in the NICE guidelines manual (NICE 00). The full scope is shown in Appendix. During the consultation period, the scope was posted on the NICE website ( Comments were invited from registered stakeholder organisations and the NICE Guideline Review Panel (GRP). Further information about the GRP can also be found on the NICE website. The NCC-C and NICE reviewed the scope in light of comments received, and the revised scope was reviewed by the GRP, signed off by NICE and posted on the NICE website. The Guideline Development Group (GDG) The ovarian cancer GDG was recruited in line with the NICE guidelines manual (NICE 00). The first step was to appoint a Chair and a Lead Clinician. Advertisements were placed for both posts and candidates were interviewed before being offered the role. The NCC-C Director, GDG Chair and Lead Clinician identified a list of specialties that needed to be represented on the GDG. Requests for applications were sent to the main stakeholder organisations, cancer networks and patient organisations/charities (see Appendix.). Individual GDG members were selected by the NCC-C Director, GDG Chair and Lead Clinician, based on their application forms. The guideline development process was supported by staff from the NCC-C, who undertook the clinical and health economics literature searches, reviewed and presented the evidence to the GDG, managed the process and contributed to drafting the guideline. At the start of the guideline development process all GDG members interests were recorded on a standard declaration form that covered consultancies, fee-paid work, share-holdings, fellowships and support from the healthcare industry. At all subsequent GDG meetings, members declared new, arising conflicts of interest which were always recorded (see Appendix.). Guideline Development Group Meetings Eleven GDG meetings were held between April 00 and 0 July 00. During each GDG meeting (either held over one or two days) clinical questions and clinical and economic evidence were reviewed, assessed and recommendations formulated. At each meeting patient/carer and service-user concerns were routinely discussed as part of a standing agenda item. NCC-C project managers divided the GDG workload by allocating specific clinical questions, relevant to their area of clinical practice, to small sub-groups of the GDG in order to simplify and speed up the guideline development process. These groups considered the evidence, as reviewed by the researcher, and synthesised it into draft recommendations before presenting it to the GDG as a whole. Each clinical question was led by a GDG member with expert knowledge of the clinical area (usually one of the healthcare professionals). The GDG subgroups often helped refine the clinical questions and the clinical definitions of treatments. They also assisted the NCC-C team in drafting the section of the guideline relevant to their specific topic. Patient/Carer Members Individuals with direct experience of ovarian cancer gave an important user focus to the GDG and the guideline development process. The GDG included three patient/carer members. They contributed as full GDG members to writing the clinical questions, helping to ensure that the evidence addressed their views and preferences, highlighting sensitive issues and terminology relevant to the guideline and bringing service-user research to the attention of the GDG. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 0 of

11 Developing Clinical Evidence-Based Questions Background Clinical guidelines should be aimed at improving clinical practice and should avoid ending up as evidence-based textbooks or making recommendations on topics where there is already agreed clinical practice. Therefore the list of key clinical issues listed in the scope were developed in areas that were known to be controversial or uncertain, where there was identifiable practice variation, or where NICE guidelines were likely to have most impact. Method From each of the key clinical issues identified in the scope the GDG formulated a clinical question. For clinical questions about interventions, the PICO framework was used. This structured approach divides each question into four components: the population (the population under study P), the interventions (what is being done - I), the comparisons (other main treatment options C) and the outcomes (the measures of how effective the interventions have been O). Where appropriate, the clinical questions were refined once the evidence had been searched and, where necessary, sub-questions were generated. The final list of clinical questions can be found in Appendix. Review of Clinical Literature Scoping search An initial scoping search for published guidelines, systematic reviews, economic evaluations and ongoing research was carried out on the following databases or websites: National Library for Health (NLH) Guidelines Finder (now NHS Evidence), National Guidelines Clearinghouse, Cochrane Database of Systematic Reviews (CDSR), Heath Technology Assessment Database (HTA), NHS Economic Evaluations Database (NHSEED), DH Data, Medline and Embase. At the beginning of the development phase, initial scoping searches were carried out to identify any relevant guidelines (local, national or international) produced by other groups or institutions. Additionally, stakeholder organisations were invited to submit evidence for consideration by the GDG, provided it was relevant to the agreed list of clinical questions. Developing the review protocol For each clinical question, the information specialist and researcher (with input from other technical team and GDG members) prepared a review protocol. This protocol explains how the review was to be carried out (see Table A) in order to develop a plan of how to review the evidence, limit the introduction of bias and for the purposes of reproducibility. All review protocols can be in the full evidence review. Table A Components of the review protocol Component Clinical question Objectives Criteria for considering studies for the review Description The clinical question as agreed by the GDG. Short description; for example To estimate the effects and cost effectiveness of or To estimate the diagnostic accuracy of. Using the PICO (population, intervention, comparison and outcome) framework. Including the study designs selected. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

12 How the information will be searched The sources to be searched and any limits that will be applied to the search strategies; for example, publication date, study design, language. (Searches should not necessarily be restricted to RCTs.) The review strategy The methods that will be used to review the evidence, outlining exceptions and subgroups. Indicate if metaanalysis will be used. Searching for the evidence In order to answer each question the NCC-C information specialist developed a search strategy to identify relevant published evidence for both clinical and cost effectiveness. Key words and terms for the search were agreed in collaboration with the GDG. When required, the health economist searched for supplementary papers to inform detailed health economic work (see section on Incorporating Health Economic Evidence ). Papers that were published or accepted for publication in peer-reviewed journals were considered as evidence. Search filters, such as those to identify systematic reviews (SRs) and randomised controlled trials (RCTs) were applied to the search strategies when there was a wealth of these types of studies. No language restrictions were applied to the search; however, foreign language papers were not requested or reviewed (unless of particular importance to that question). The following databases were included in the literature search: The Cochrane Library Medline and Premedline 0 onwards Excerpta Medica (Embase) 0 onwards Cumulative Index to Nursing and Allied Health Literature (Cinahl) onwards Allied & Complementary Medicine (AMED) onwards British Nursing Index (BNI) onwards Psychinfo 0 onwards Web of Science [specifically Science Citation Index Expanded] (SCI-EXPANDED) onwards and Social Sciences Citation Index (SSCI) onwards] Biomed Central onwards From this list the information specialist sifted and removed any irrelevant material based on the title or abstract before passing to the researcher. All the remaining articles were then stored in a Reference Manager electronic library. Searches were updated and re-run weeks before the stakeholder consultation, thereby ensuring that the latest relevant published evidence was included in the database. Any evidence published after this date was not included. For the purposes of updating this guideline, July 00 should be considered the starting point for searching for new evidence. Further details of the search strategies, including the methodological filters used, are provided in the evidence review (and appear on the CD-ROM accompanying this guideline). Critical Appraisal From the literature search results database, one researcher scanned the titles and abstracts of every article for each question and full publications were ordered for any studies considered relevant or if there was insufficient information from the title and abstract to inform a decision. When the papers were obtained the researcher applied The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

13 0 0 0 inclusion/exclusion criteria to select appropriate studies which were then critically appraised. For each question, data on the type of population, intervention, comparator and outcomes (PICO) were extracted and recorded in evidence tables and an accompanying evidence summary prepared for the GDG (see evidence review). All evidence was considered carefully by the GDG for accuracy and completeness. GRADE (Grading of Recommendations, Assessment, Development and Evaluation) For interventional questions, studies which matched the inclusion criteria were evaluated and presented using a modification of GRADE (NICE 00; group.org/). Where possible this included meta-analysis and synthesis of data into a GRADE evidence profile. The evidence profile shows, for each outcome, an overall assessment of both the quality of the evidence as a whole (low, moderate or high) as well as an estimate of the size of effect. A narrative summary (evidence statement) was also prepared. Each topic outcome was examined for the quality elements defined in table B and subsequently graded using the quality levels listed in table C. The reasons for downgrading or upgrading specific outcomes were explained in footnotes. Table B Descriptions of quality elements of GRADE Quality element Limitations Inconsistency Indirectness Imprecision Publication bias Description Limitations in the study design and implementation may bias the estimates of the treatment effect. Major limitations in studies decrease the confidence in the estimate of the effect. Inconsistency refers to an unexplained heterogeneity of results. Indirectness refers to differences in study population, intervention, comparator or outcomes between the available evidence and the clinical question. Results are imprecise when studies include relatively few patients and few events and thus have wide confidence intervals around the estimate of the effect relative to the minimal important difference. Publication bias is a systematic underestimate or overestimate of the underlying beneficial or harmful effect due to the selective publication of studies. Table C Overall quality of outcome evidence in GRADE Quality element High Moderate Low Very low Description Further research is very unlikely to change our confidence in the estimate of effect. Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Any estimate of effect is very uncertain. All procedures were fully compliant with NICE methodology as detailed in the NICE guidelines manual (NICE 00). In general, no formal contact was made with authors; however, there were ad hoc occasions when this was required in order to clarify specific details. Needs Assessment As part of the guideline development process the NCC-C invited a specialist registrar, with the support of the GDG, to undertake a needs assessment (see Appendix.). The needs The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

14 assessment aims to describe the burden of disease and current service provision for patients with ovarian cancer in England and Wales, which informed the development of the guideline. Assessment of the effectiveness of interventions is not included in the needs assessment, and was undertaken separately by researchers in the NCC-C as part of the guideline development process. The information included in the needs assessment document was presented to the GDG. Most of the information was presented in the early stages of guideline development, and other information was included to meet the evolving information needs of the GDG during the course of guideline development. Incorporating Health Economics Evidence The aim of providing economic input into the development of the guideline was to inform the GDG of potential economic issues relating to the recognition and initial management of ovarian cancer. It is important to investigate whether health services are both clinically effective and cost effective, i.e. are they value for money. Prioritising topics for economic analysis In addition to the review of the relevant clinical evidence, the GDG were required to determine whether or not the cost-effectiveness of each of the individual clinical questions should or could be investigated. After the clinical questions were decided, and with the help of the health economist, the GDG agreed which of the clinical questions were an economic priority for analysis. Further details of the economic prioritisation are provided in the evidence review (and appear on the CD-ROM accompanying this guideline). These economic priorities were chosen on the basis of the following criteria, in broad accordance with the NICE guidelines manual (NICE 00): Overall relevance of the topic: The number of patients affected: interventions affecting relatively large numbers of patients were given a higher economic priority than those affecting fewer patients The health benefits to the patient: interventions that were considered to have a potentially significant impact on both survival and quality of life were given a higher economic priority The per patient cost: interventions with potentially high financial (cost/savings) implications were given high priority compared to interventions expected to have lower financial implications Likelihood of changing clinical practice: priority was given to topics that were considered likely to represent a significant change to existing clinical practice. Uncertainty: High level of existing uncertainty: higher economic priority was given to clinical questions in which further economic analysis was considered likely to reduce current uncertainty over cost-effectiveness. Low priority was given to clinical questions when the current literature implied a clearly attractive or unattractive incremental costeffectiveness ratio, which was regarded as generalisable to a UK healthcare setting Likelihood of reducing uncertainty with further analyses (feasibility issues): when there was poor evidence for the clinical effectiveness of an intervention, there was considered to be less justification for an economic analysis to be undertaken. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

15 For each topic that was prioritised for economic analysis a comprehensive systematic review of the economic literature was conducted. Where published economic evaluation studies were identified that addressed the economic issues for a clinical question, these are presented alongside the clinical evidence wherever possible. For those clinical areas reviewed, the information specialists used a similar search strategy as used for the review of clinical evidence but with the inclusion of a health economics filter. Each search strategy was designed to find any applied study estimating the cost or cost effectiveness of the topic under consideration. A health economist reviewed abstracts and relevant papers were ordered for appraisal. Published economic evidence was obtained from a variety of sources: Cochrane HTA NHS Economic Evaluations Database (NHS EED) Medline Embase. Economic Analysis Once the priority topics for economic analysis had been agreed by the GDG, the health economist investigated whether or not a cost-effectiveness analysis of each topic could be carried out. Cost-effectiveness evaluations require evidence on numerous parameters, including treatment effects, health-related preferences (utilities), healthcare resource use and costs. However, high quality evidence on all relevant parameters within an economic analysis is not always available. If the evidence base used to inform a cost-effectiveness analysis is poor, decisions based upon such an analysis may be subject to a high degree of uncertainty and therefore cost effectiveness analysis would not be appropriate. For those clinical questions where an economic model was required, cost-utility analysis was undertaken using a decision tree approach. Decision tree is an analytical method of evaluating all options and consequences relevant to a specific decision problem. Assumptions and designs of the decision models were explained to and agreed by the GDG members during meetings, and they commented on subsequent revisions. The details of the model are presented in the evidence review and Appendix. During the analysis the following general principles were adhered to: the GDG Chair and Clinical Lead were consulted during the construction and interpretation of the analysis the analysis was based on the best evidence from the systematic review assumptions were reported fully and transparently the results were subject to thorough sensitivity analysis and limitations discussed costs were calculated from a health services perspective. Linking to NICE technology appraisals When this guideline was commissioned there was one published technology appraisal (TA) which was relevant to the guideline (TA: Paclitaxel for the treatment of ovarian cancer; Published TAs are periodically reviewed to determine if they need to be updated particularly if any new evidence becomes available since the publication of the appraisal which means the original recommendations needed to be changed. In October 00, NICE consulted with stakeholders to assess whether TA should be updated within the guideline. The outcome was that TA should remain on the static list and therefore its recommendations were reproduced unchanged in the most appropriate section of the guideline The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

16 Agreeing the Recommendations For each clinical question the GDG were presented with a summary of the clinical evidence, and where appropriate economic evidence, derived from the studies reviewed and appraised. From this information the GDG were able to derive the guideline recommendations. The link between the evidence and the view of the GDG in making each recommendation is made explicit in the accompanying LETR statement. LETR (Linking Evidence to Recommendations) statements As clinical guidelines were previously formatted, there was limited scope for expressing how and why a GDG made a particular recommendation from the evidence of clinical and cost effectiveness. To make this process more transparent to the reader, NICE have introduced an explicit, easily understood and consistent way of expressing the reasons for making each recommendation. This is known as the LETR statement and will usually cover the following key points: the relative value placed on the outcomes considered the strength of evidence about benefits and harms for the intervention being considered the costs and cost-effectiveness of an intervention (if formally assessed by the health economics team) the quality of the evidence (see GRADE) the degree of consensus within the GDG other considerations for example equalities issues Where evidence was weak or lacking the GDG agreed the final recommendations through informal consensus. Shortly before the consultation period, ten key priorities and five key research recommendations were selected by the GDG for implementation and the patient algorithms were agreed. To avoid giving the impression that higher grade recommendations are of higher priority for implementation, NICE no longer assigns grades to recommendations. Consultation and Validation of the Guideline The draft of the guideline was prepared by NCC-C staff in partnership with the GDG Chair and Lead Clinician. This was then discussed and agreed with the GDG and subsequently forwarded to NICE for consultation with stakeholders. Registered stakeholders (see Appendix.) had one opportunity to comment on the draft guideline which was posted on the NICE website between September 00 and November 00 in line with NICE methodology (NICE 00). The Guideline Review Panel also reviewed the guideline and checked that stakeholder comments had been addressed. The pre-publication check process Following stakeholder consultation and subsequent revision, the draft guideline was then subject to a pre-publication check (NICE 00). The pre-publication check provides registered stakeholders with the opportunity to raise any concerns about factual errors and inaccuracies that may exist in the revised guideline after consultation. During the pre-publication check the full guideline was posted on the NICE website for working days, together with the guideline consultation table that listed comments received during consultation from stakeholders and responses from the NCC-C and GDG. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

17 All stakeholders were invited to report factual errors using a standard proforma. NICE, the NCC and the GDG Chair and Lead Clinician considered the reported errors and responded only to those related to factual errors. A list of all corrected errors and the revised guideline were submitted to NICE, and the revised guideline was then signed off by Guidance Executive. The list of reported errors from the pre-publication check and the responses from the NCC-C were subsequently published on the NICE website. The final document was then submitted to NICE for publication on their website. The other versions of the guideline (see below) were also discussed and approved by the GDG and published at the same time. Other Versions of the Guideline This full version of the guideline is available to download free of charge from the NICE website ( and the NCC-C website ( NICE also produces three other versions of the ovarian cancer guideline which are available from the NICE website: the NICE guideline, which is a shorter version of this guideline, containing the key priorities, key research recommendations and all other recommendations the Quick Reference Guide (QRG), which is a summary of the main recommendations in the NICE guideline. For printed copies, phone NICE publications on 0 00 or publications@nice.org.uk Understanding NICE Guidance ( UNG ), which describes the guideline using nontechnical language. It is written chiefly for people suspected of, or diagnosed with, ovarian cancer but may also be useful for family members, advocates or those who care for patients with cancer of unknown primary. For printed copies, phone NICE publications on 0 00 or publications@nice.org.uk Updating the Guideline Literature searches were repeated for all of the clinical questions at the end of the GDG development process, allowing any relevant papers published before July 00 to be considered. Future guideline updates will consider evidence published after this cut-off date. Three years after publication of the guideline, NICE will commission a National Collaborating Centre to determine whether the evidence base has progressed significantly to alter the guideline recommendations and warrant an early update. Funding The National Collaborating Centre for Cancer was commissioned by NICE to develop this guideline. Health economic analysis for this guideline was provided by the London School of Hygiene and Tropical Medicine and funded by the National Collaborating Centre for Cancer. Disclaimer The GDG assumes that healthcare professionals will use clinical judgment, knowledge and expertise when deciding whether it is appropriate to apply these guidelines. The recommendations cited here are a guide and may not be appropriate for use in all situations. The decision to adopt any of the recommendations cited here must be made by the practitioner in light of individual patient circumstances, the wishes of the patient and clinical expertise. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

18 0 The NCC-C disclaims any responsibility for damages arising out of the use or non-use of these guidelines and the literature used in support of these guidelines. References Briggs, A., Claxton K, Sculpher M, Decision Modelling for Health Economic Evaluation. 00, Oxford: Oxford University Press. National Institute for Health and Clinical Excellence (00) The guidelines manual. London: National Institute for Health and Clinical Excellence. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

19 Algorithm Woman accessing healthcare Patient support Identification of a pelvic mass and/or ascites on physical examination New diagnosis of IBS in a women > 0 years of age Ultrasound of the abdomen and pelvis Measure serum CA and perform ultrasound of the abdomen and pelvis if not already done. Measure AFP and beta-hcg in women < 0 years of age. Reports persistent or frequent abdominal distension/bloating, difficulty eating and/or feeling full, pelvic or abdominal pain, increased urinary urgency and/or frequency. Especially in woman > 0 years of age Measure serum CA Urgent referral Advise the woman to return if her symptoms persist or increase in frequency Risk of Malignancy Index I 00 Refer to specialist MDT. This should include access to information as individually required on: stage of disease, treatment options, management of side effects, sex and sexuality, fertility and hormonal treatment, disease recurrence, self-help, genetic risks, available support, managing emotions. Suspected early stage (stage I) ovarian cancer Above IU/ml Suggests ovarian cancer Normal IU/ml or below CT scan of the pelvis, abdomen and thorax (do not routinely use MRI) Consider tests if: abnormal vaginal bleeding, unexplained weight loss, abdominal distension, fatigue, or change in bowel habit Not tested Suspected advanced (stage II- IV) ovarian cancer Primary care Secondary care Surgical staging. Do not perform systematic retroperitoneal lymphadenectomy if cancer confined to ovaries. Confirm diagnosis using surgical specimen or biopsy not cytology. Use percutaneous imageguidedbiopsy if feasible; only use laparoscopy for biopsy when not. Only offer cytotoxic chemotherapy without a tissue diagnosis in exceptional circumstances where risks have been discussed with the woman and MDT Optimally staged as Ia/b (grade - ): no chemotherapy Sub-optimally staged as I: discuss risks and benefits of chemotherapy with the women Stage Ia/b (grade ), stage Ic and above: offer cycles of carboplatin or consider cycles of carboplatin plus paclitaxel Do not offer intraperitoneal chemotherapy except as part of a clinical trial This algorithm summarises the recommendations made in this guideline but is not to be regarded as a comprehensive clinical pathway in the management of ovarian cancer. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

20 Epidemiology. Introduction This chapter provides a summary of the needs assessment that was carried out to inform development of this guideline and includes current information regarding the epidemiology of ovarian cancer.. Data collection Office of National Statistics (ONS) and cancer registries The data on incidence, mortality and survival of ovarian cancer for the United Kingdom is published by the ONS (00). It is based on the data collated by cancer registries covering England, Wales, Scotland and Northern Ireland (Department of Health, 00). Sources for this data include general hospitals, cancer centres, hospices, private hospitals, cancer screening programmes, primary care, nursing homes and death certificates. The minimum dataset of information includes: Patient details (name, date of birth, NHS number, address, ethnicity and sex) Hospital details (hospital, consultant and patient unit number) Diagnostic, tumour and treatment details (site and type of primary tumour, laterality, stage and grade of the tumour, and some treatment information) Death details (date of death, cause and place of death and post mortem information). There is approximately a two year gap between the event time and the publication of the summary statistics. There is a high degree of completeness in terms of diagnosis and deaths. However, the completeness and quality of data collected on a specific individual and their cancer can be variable. Registries record information about cancers apparent at the time of diagnosis of the primary neoplasm. However, they do not always record information about management and treatment received. Consequently national data on the management of ovarian cancer is sparse Some international data are available from GLOBOCAN and EUROCARE and are valuable for the purpose of comparison. The GLOBOCAN project provides contemporary estimates of the incidence of, and mortality from the major types of cancer at a national level, for all countries of the world. The GLOBOCAN estimates are presented for 00 separately by sex and for all ages. These are calculated from the recent data provided by the International Agency of Research for Cancer (IARC). The EUROCARE project seeks to standardise the cancer survival data across Europe in order to provide meaningful comparisons between countries (Berrino, 00). An important point to remember when looking at the results is that cancer registration in several European countries only covers a small proportion of the total national population. Summary results for these countries may not therefore represent the situation in the country as a whole and hence might not be a true comparison (Berrino et al., 00). The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 0 of

21 0 0 0 Hospital inpatient care In England, the Hospital Episode Statistics (HES) record information on all NHS admissions. These include all day case and inpatient admissions to NHS hospitals (including private patients and non-uk residents) and admissions to independent providers commissioned by the NHS. The information recorded includes patient demographic information, diagnosis for each admission and date and length of admission. A similar system, Patient Episode Database Wales (PEDW) operates in Wales. The data is processed nationally to remove duplicates and any obvious errors in order to provide the most robust data possible. The quality of the data is only as good as the quality of data entry and this may vary between providers. Systematic misclassification will occur but it is not possible to quantify and its effect is unknown. The Welsh Cancer Intelligence and Surveillance Unit (WCISU) has combined their registry and HES/PEDW data to obtain information on the treatment received by ovarian cancer patients in their locality. There is a similar project being carried out in England by the Trent Cancer Registry and the results are expected later this year. Hospital outpatient care Outpatient data have also been collected through the HES and PEDW dataset since 00. These data record the speciality associated with the appointment but does not record the particular investigation carried out or the results of the appointment and so have not been examined as part of this needs assessment.. Incidence Ovarian cancer is the fifth commonest cancer in women in the UK after breast, colorectal, lung and uterus. Approximately,00 new cases of ovarian cancer were diagnosed every year in United Kingdom between 00 and 00 accounting for approximately in 0 cases of cancer in women (Walsh and Cooper, 00). Incidence in the UK, constituent countries and cancer networks Data in Table. show that in 00, new cases of ovarian cancer were diagnosed in the UK which equates to a crude rate of. per 00,000 population. The European age standardised incidence rate (EASR) is.0 per 00,000 population. There are slight variations in the incidence rate across the constituent countries of the UK. Wales has a higher incidence rate compared to the national rates and Northern Ireland the lowest (. per 00,000 population). Table. Number of new cases and rates registered for ovarian cancer in 00. England Wales Scotland N.Ireland United Kingdom Cases,, Crude rate per 00,000 population Age-standardised rate (European) per 00,000 population % CI Data source: Reproduced from Cancer Research UK. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

22 Age-standarised Incidence rates per 00,000 DRAFT FOR CONSULTATION 0 The latest data of incidence rate by cancer network in England and Wales is from 00 (Figure.). Comparing networks within England, the incidence rate was highest in the Peninsula Cancer Network with a rate of. per 00,000 population. The lowest incidence rate was noted in the North East London Cancer Network with an incidence rate of.0. All cancer networks in Wales had rates higher than the UK average. These differences in the incidence rates across the UK may have arisen from differences in diagnostic criteria or cancer registration or both. Figure. Age-standardised incidence rates of ovarian cancer by Cancer Network in England and Wales (00) North of Scotland South West London Counties Avon, Somerset and Wiltshire Merseyside and Cheshire South East London Yorkshire Mount Vernon Scotland Kent and Medway United Kingdom West of Scotland Dorset North Trent Surrey, West Sussex and Hampshire Pan Birmingham Northern Ireland South East of Scotland South East Wales Humber and Yorkshire Coast West London Central South Coast Lanchashire and South cumbria North Wales Essex Thames Valley Wales Mid Trent Sussex North East London Greater Midlands England South West Wales Greater Manchester and Cheshire Leics, Northants and Rutland Penninsula North of England Derby/Burton Arden Anglia North London Incidence Rate Data sources: ISD Scotland, Northern Ireland Cancer Registry, UK Association of Cancer Registries, Welsh Cancer Intelligence and Surveillance These data include borderline malignancies. A further confounding issue is that primary peritoneal cancer and metastatic malignant disease of unknown primary origin may also be included. European and Worldwide comparison Figure. shows the incidence rates of ovarian cancer across the world in 00. The United Kingdom has a relatively high incidence rate of up to. per 00,000 population. The incidence rates are highest in Central America and Northern Europe and lowest in some parts of Africa and Asia. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

23 Figure. Worldwide estimated age-standardised incidence rate of ovarian cancer per 00,000 population; all ages (00) 0 Data source: Globocan 00(IARC) In comparison with other European countries, the UK is among those with the highest incidence rates of ovarian cancer (Figure.). Generally the highest rates are in the Northern and Eastern European countries of Lithuania, Latvia, Ireland, Slovakia and Czech Republic. The lowest rates are in Southern European countries of Portugal and Cyprus. Figure. Age-standardised incidence rates of ovarian cancer in the European Union (00). United Kingdom The Netherlands Sweden Spain Slovenia Slovakia Portugal Poland Malta Luxembourg Lithuania Latvia Italy Ireland Hungary Greece Germany France Finland Estonia Denmark Czech Republic Cyprus Belgium Austria Incidence Data source: Globocan 00 (IARC) Rate per 00,000 females The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

24 Number of cases Rate per 00,000 population DRAFT FOR CONSULTATION Incidence rates of ovarian cancer by age The lifetime risk of women being diagnosed with ovarian cancer is in (Walsh and Cooper, 00). The data in Figure. show that overall 0% of the ovarian cancer recorded in the UK in 00 were in women aged years and above. The incidence rates are higher in postmenopausal women, with the highest in the age group of 0- years of age. Figure. Number of new cases diagnosed and incidence rate of ovarian cancer by age in the United Kingdom (00).,000 Female cases Female rates Age at diagnosis Data source: Office for National Statistics. Trends in incidence rates of ovarian cancer The age standardised incidence rates of ovarian cancer have increased in the UK from. in to. in 00 (Figure.). Incidence rates peaked around - and this may be associated with the inclusion of cancer of borderline malignancy within the category of malignant cancer according to International Classification of Disease for Oncology (ICDO). The ICDO was introduced in England and Wales in, Scotland in and Northern Ireland in. This could also explain the rising trend of incidence rates after. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

25 A ge Standarised Incidence R ate DRAFT FOR CONSULTATION Figure. Trends in age standardised incidence rates of ovarian cancer (-00) Incidence Data Source: Cancer Research UK Socioeconomic status and ethnicity Socioeconomic status has no affect on incidence of ovarian cancer (Figure.). Year The National Cancer Intelligence Network (NCIN) recently published a report analysing the relationship between cancer incidence and ethnicity in those diagnosed with cancer in England (00-00) (NCIN, 00). It showed Asian and Black ethnic groups have lower incidence rates of ovarian cancer compared to the White ethnic group. The analysis was presented only on Asian, Black and White ethnic group due to the small number of Chinese and Mixed ethnic groups in the study. Figure. Ovarian cancer incidence by deprivation quintile, England (000-00) Least Deprived Most Deprived Overall Data source: NCIN 00.. Mortality Approximately,00 women die from ovarian cancer each year in the UK which makes it the leading cause of death in gynaecological cancers (Cancer Research UK). It accounts for % of all cancer deaths in women. The reason for the high mortality rate in ovarian cancer may be because most women are diagnosed with advanced ovarian cancer at the time of detection. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

26 Age-standarised Mortality rate per 00,000 population DRAFT FOR CONSULTATION 0 0 Mortality rates in the United Kingdom The age-standardised mortality rates are similar across all countries within the UK with an overall average of. (Table.). The highest mortality rate is seen in Northern Ireland (.0) compared to the UK average. Wales has the lowest mortality rate in spite of a high incidence rate (see Table.). Table. Number of deaths and European age-standardised mortality rates of ovarian cancer per 00,000 population in the UK (00) England Wales Scotland N. Ireland United Kingdom Deaths,0, Crude rate per 00,000 population Age-standardised rate (European) per 00,000 population % CI Data source: Office of National Statistics, reproduced from Cancer Research UK Mortality rates by cancer network The mortality rate of ovarian cancer by cancer network in 00 was highest in the Peninsula and Mid Trent Cancer Network and lowest in the North London, West London and North East London Cancer Networks (Figure.). Figure. Age-standardised mortality rates of ovarian cancer by cancer network in the UK (00) 0 0 North London CN West London CN North East London CN South West London CN Yorkshire CN South West Wales CN West of Scotland CN Thames Valley CN North of England CN Avon, Somerset and Wiltshire CN Greater Manchester and Cheshire CN Scotland Kent and Medway CN North of Scotland CN Counties CN South East London CN Dorset CN Anglia CN Greater Midlands CN England United Kingdom North Trent CN Leics, Northants and Rutland CN Merseyside and Cheshire CN Humber and Yorkshire Coast CN Mount Vernon CN South East of Scotland CN Central South Coast CN Surrey, West Sussex and Hampshire CN Pan Birmingham CN Wales North Wales CN Lanchashire and South cumbria Sussex CN Arden CN Derby/Burton CN Northern Ireland Essex CN Penninsula CN Mid Trent CN South East Wales CN Data sources: ISD Scotland; Northern Ireland Cancer Registry; UK Association of Cancer Registries; Welsh Cancer Intelligence and Surveillance Unit; NCIN 00 Mortality rates and number of deaths by age Data in Figure. show the number of deaths and mortality rate by age in the UK in 00. The number of deaths is highest in 0- years age group, but the highest mortality rates are in the 0- years age group. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

27 Number of deaths Age-specific rates per women 00,000 DRAFT FOR CONSULTATION Figure. Number of deaths and mortality rate of ovarian cancer in the UK by age (00) Age Data source: Reproduced from Cancer Research UK. Worldwide and European comparisons The global and European data in this section for ovarian cancer are contemporary estimates from the GLOBOCAN project (Figure.). The advantage of global data is national coverage and long-term availability. However, the data quality varies considerably. These data indicate that the United Kingdom and Ireland have comparatively high mortality rates even when compared to other European countries. Figure. Worldwide estimated age-standardised mortality rate of ovarian cancer per 00,000 population, all ages (00) 0 Data source: Globocan 00 (IARC) Across Europe, the highest mortality rates are seen in Northern Europe and Ireland (Figure.0). This is similar to the high incidence rates seen in these regions. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

28 Rate per 00,000 population DRAFT FOR CONSULTATION Figure.0 Age-standardised mortality rate of ovarian cancer, European Union (00) United Kingdom The Sweden Spain Slovenia Slovakia Portugal Poland Malta Luxembourg Lithuania Latvia Italy Ireland Hungary Greece Germany France Finland Estonia Denmark Czech Republic Cyprus Belgium Austria Age -standardised mortality rates 0 Data source: Globocan 00 (IARC) Trends in mortality rates and numbers of deaths from ovarian cancer Data in Figure. show the trends in the age-specific mortality rate of ovarian cancer from to 00. The trends vary across the different age groups. The mortality rate shows a gradual increase in women over years of age with some decline in younger women. It is evident from the graph that the mortality rate has been fairly stable over the last 0 years in women under years of age compared to the age group of 0- years where there has been a steady decline. Overall mortality rate of ovarian cancer remains relatively stable in spite of the increasing incidence. Figure. Trends in age specific mortality rate of ovarian cancer by age in United Kingdom (-00) Overall mortality Data source: Reproduced from Cancer Research UK. Year of death The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

29 Relative Survival Rate DRAFT FOR CONSULTATION 0. Survival Most women are diagnosed with advanced stage disease and consequently ovarian cancer has the lowest relative five year survival rate of all gynaecological cancers (ONS 00). Trends in survival rates from ovarian cancer The five year survival rates for patients with ovarian cancer have increased dramatically from 0% in to.% in 00 (Figure.). A similar trend has been observed in ten year survival rate from 0% between - to.% between -000 (Figure.). The two fold increase in the survival rate may be due to early detection methods, improved treatment modalities, or inclusion of borderline tumours which have a good prognosis (ONS 00; Richard 00; Rachet et al., 00). Figure. Trends in the age-standardised one year, five year and ten year(-000) survival rate of ovarian cancer in England and Wales (-00) year- year- year * 00-00** (* England only data, ** shows one year survival between and five year survival between 00-00) Data source: Office of National Statistics and Cancer Research UK Survival rate by age at diagnosis The survival rate based on age at diagnosis is shown in Figure.. Both the one-year and five year survival are higher in young women (-) compared to older women (>0). In women aged - years the one year and five year survival are % and % respectively compared to % and % in the 0- age group. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

30 Relative survival rate, Cumulative Age standardised relative survival rate DRAFT FOR CONSULTATION Figure. Age-standardised five year relative survival of ovarian cancer by age in England (00-00) Five-year survival Age Data source: Office of National Statistics-Statistical Bulletin Cancer survival in England (Berrino 00; Berrino et al., 00) European comparison In an international comparison of women diagnosed with ovarian cancer in -, the survival rates in England, Wales, Scotland and Northern Ireland were significantly lower than the European average (Figure.). Figure. Relative five year survival rate, cumulative of ovarian cancer for women aged - years diagnosed - across Europe Austria Belgium Czech Republic Data source: Eurocare Database Denmark Finland France Germany Iceland Ireland Italy Malta Netherlands Norway Poland Portugal Slovenia Spain Sweden Switzerland UK England UK Northern Ireland UK Scotland UK Wales Slovakia Europe The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 0 of

31 0 0 0 Survival by stage Ovarian cancer is staged using the FIGO classification (Box.), based on the information obtained from surgery, supplemented by imaging information where appropriate. Cancer registries use TNM classification similar to FIGO staging. Box. FIGO staging for ovarian cancer Stage I: limited to one or both ovaries Ia involves one ovary; capsule intact; no tumour on ovarian surface; no malignant cells in ascites or peritoneal washings b involves both ovaries; capsule intact; no tumour on ovarian surface; negative washings c tumour limited to ovaries with any of the following: capsule ruptured, tumour on ovarian surface, positive washings Stage II: pelvic extension or implants IIa extension or implants onto uterus or fallopian tube; negative washings IIb extension or implants onto other pelvic structures; negative washings IIc pelvic extension of implants with positive peritoneal washings Stage III: microscopic peritoneal implants outside of the pelvis; or limited to the pelvis with extension to the small bowel or omentum IIIa microscopic peritoneal metastases beyond pelvis IIIb macroscopic peritoneal metastases beyond pelvis less than cm in size IIIc peritoneal metastases beyond pelvis > cm or lymph node metastases Stage IV: distant metastases to the liver or outside the peritoneal cavity Currently there is only data available in Wales on the stage at presentation for women with ovarian cancer. Data from WCISU showed that only 0-0% of staging data are recorded on their Cancer registry database for patients with ovarian cancer (Figure.). This makes statistical analysis based on staging difficult. Data from England is expected to be published in September 00. Figure. Ovarian cancer by stage, Wales (000-00) Year Stage Stage Stage Stage Unknown 0% 0% 0% 0% 0% 00% Data source: WCISU Number The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

32 0 0 0 Socioeconomic status and ethnicity Among adults living in the most deprived areas who were diagnosed cancer between and 0, -year survival was significantly lower than for those in the most affluent areas for of different cancers (Coleman et al., ).. Routes to diagnosis Regional data was obtained from Avon, Somerset and Wiltshire Cancer Network on routes to diagnosis. This was undertaken as an initial study to find out how people came to be diagnosed with cancer. The results for ovarian cancer according to age group are shown in Table.. These data indicate that the vast majority of patients attend electively rather than as emergencies. A significant proportion of elective admissions present outside the urgent (two week) referral pathway. Table. Routes to diagnosis for ovarian cancer by age group Age (years) GP/OP referral Two week wait Emergency presentation Other OPD* Inpatient elective Unknown Death Certificate only - 0% 0% 00% 0% 0% 0% 0% 0-0% % % 0% 0% 0% 0% - 0% % % % 0% % 0% 0- % 0% 0% % % % 0% - % % % % % 0% 0% 0-0% 0% 0% 0% 0% 0% 0% - % % % 0% % % 0% 0- % 0% % % % % 0% - % % % % % 0% 0% 0- % % % % % % 0% - % % % % % % 0% 0- % % % % % % % - 0% % % % % % 0% 0- % % % % % % 0% - % % % % % % 0% 0-0% % % % % % % & above % % % % % 0% % * all other outpatient appointments other than gynaecological two week wait women admitted for elective procedures Data source: reproduced with permission from Lucy Elliss-Brookes; Avon, Somerset and Wiltshire Cancer Network (Elliss- Brookes, 00).. Treatment Ovarian cancer is managed using a number of treatments which usually comprise chemotherapy or surgery often in combination. As there was no available comparative national data on treatment modalities, a questionnaire was developed by the GDG and sent to all cancer networks. Only two cancer networks were able to provide data on treatments used. In one region it appeared that up to 0% of patients are managed with chemotherapy alone (this had an association with age). In the other region there was marked variation between hospitals and within hospitals over time in the proportion of patients receiving chemotherapy. The reason for this variation is not understood. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

33 Su rg ic a l Pro c e d u re s DRAFT FOR CONSULTATION 0 Surgery Currently there is only data available in Wales on the surgical management of women with ovarian cancer. Data from England is expected to be published in September 00. WCISU recently combined PEDW data on the surgical management of women with ovarian cancer using data from the financial years 00 to 00. There were a total of women diagnosed with ovarian cancer during that time. Figure. illustrates the different procedures carried out in the three cancer networks in Wales. The most frequent procedure undertaken involves total abdominal hysterectomy, bilateral salphingo-oopherectomy and omentectomy as this involves the staging laparotomy. Figure. Number of different surgical procedures performed for ovarian cancer by cancer network, Wales (00-00) Omentectomy Ğ Supraracolic Removal of unilateral ovaries Bilateral removal of ovaries North Wales South West Wales South East wales Total abdominal hysterectomy Number of procedures Data source: WCISU. The findings of cancer peer review of gynaecology cancer teams in England The Calman-Hine report on a Policy Framework for Commissioning Cancer Services published in and the series of NICE Improving Outcome Guidance formed the basis of establishing national standards for cancer care in England. This led to the establishment of a National Cancer Peer Review (NCPR) process which is a national quality assurance programme for NHS cancer services in England. It aims to improve the care of the patients with cancer and their families. This is done through self-assessment by cancer service teams and external review by professional peers against nationally agreed quality peer review measures. The first programme of review focussed on services in four tumour site areas; breast, lung, colorectal, gynaecology and was coordinated on a regional rather than national basis. The programme was independently evaluated, the results of which informed the development of the 00-0 National Cancer Peer Review Programme. Currently the NCPR programme consists of the three key stages illustrated in the Figure.. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

34 Figure. Stages of the National Cancer Peer Review Programme on gynaecology cancer teams (00-00) All cancer networks in England and all their designated local and specialist Gynaecology cancer teams were reviewed against the national standards by a team of clinical peers between 00 and 00. The reports of these reviews are available publicly via the CQuiNS website. About local multidisciplinary teams (MDTs) were reviewed. The review was for all gynaecological cancers and not for ovarian cancer alone. During the targeted visit, the peer group reviewed whether each measure is achieved or not and whether overall progress is being made toward the achievement of the standards. Following the outcome of the review, the cancer networks should agree actions in order to meet those standards not currently being met achieved within defined timescales. The results of the most recent peer review process in England (00-00) are currently being analysed and a national report of compliance by cancer networks is expected later this year. This report will also compare results from the peer review process with the most recent data. We hope to include this in the final guideline.. Summary Ovarian cancer is the second most common gynaecological cancer in the UK accounting for over,00 new cases diagnosed each year. The rates have been steadily increasing over the past 0- years, with a notable increase in the years and above age group. There is some geographic variation in the incidence rate across the UK. This may be due to variation in diagnostic criteria, cancer registration or population. Ovarian cancer is the leading cause of death in women with gynaecological cancer and accounts for % of all deaths in women. The mortality rate remains almost the same in all regions of the UK. There has been a two fold increase in the survival rate over the last two decades which might reflect better diagnostic and treatment methods. The process of producing this report has highlighted the lack of data available to assess the burden of the disease based on the stage and the type of ovarian cancer. It is clear that The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

35 0 0 there are difficulties in the collection and definitions in the minimum dataset for ovarian cancer. This deficiency makes the interpretation of effectiveness of treatments impossible and is an important obstacle to improving cancer care for women with ovarian cancer. References ONS (00) Cancer statistics registrations: registrations of cancer diagnosed in 00. England. Series MB no.. Department of Health (00) The national cancer registration system 00. Berrino F (00) The EUROCARE study: strengths, limitation and perspectives of population bases, comparative survival studies. Annals of Oncology : - Berrino F., Verdecchia A., Lutz J.M., et al.,: the EUROCARE working group (00) Comparative cancer survival information in Europe. European Journal of Cancer : 0-0. Coleman MP., Babb P., Damiecki P., et al., () Cancer Survival Trends in England and Wales : Deprivation and NHS Region. London: The Stationery Office; Series SMPS No.. Walsh P. and Cooper N. (00) Ovary, Cancer Atlas of the United Kingdom and Ireland 000. p: -0 Office for National Statistics (ONS) (00) Survival Rates in England, patients diagnosed followed up to 00. Available at Richard MA. (00) Trends and inequalities in survival for 0 cancers in England and Wales - 00: population-based analyses and clinical commentaries..british Journal Cancer. (Supp ): S. Rachet B., Maringe C., Nur U. et al., 00. Population-based cancer survival trends in England and Wales up to 00:an assessment of the NHS cancer plan for England.The Lancet Oncology.. 0(): -. NCIN (00) Cancer incidence and survival by major ethnic group, England, 00-00: incidence rate ratios and estimated ASRs. NCIN (00) Cancer incidence and mortality by cancer network, UK, Elliss-Brookes L. (00) Routes to Diagnosis results: presentation at UKACR and NCIN joint conference. Available at 0 The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

36 Detection in primary care The challenge presented by ovarian cancer is to make the correct diagnosis as early as possible despite the non-specific nature of symptoms and signs. It is therefore important to establish those symptoms and signs which initiate the first best test that will ensure the woman is directed to the most appropriate clinical pathway. The two objectives of this chapter were:. to identify which symptoms and signs are associated with ovarian cancer to potentially allow earlier recognition of ovarian cancer in primary care. to assess the relationship between the duration of symptoms and ovarian cancer outcome.. Awareness of symptoms and signs Early recognition of ovarian cancer symptoms Ovarian cancer has been termed the silent killer but it is increasingly recognised that the majority of women with ovarian cancer have symptoms. These symptoms are non-specific and widely experienced among the general population. However, they have greater significance in older women (over 0 years of age) and in those with a significant family history (two or more cases of ovarian or breast cancer diagnosed at an early age in first degree relatives). Two important pieces of work have been published on the signs and symptoms of ovarian cancer which should be considered alongside the recommendations in this guideline. In 00 NICE published a set of recommendations for GPs for the urgent referral of woman suspected of having gynaecological cancer, including ovarian cancer (NICE, 00). This guideline updates and will replace recommendation.. in Referral guidelines for suspected cancer (NICE clinical guideline ; published June 00). NICE are currently reviewing whether the entire guideline should be updated and a decision is expected in November 00. A more recent programme has been the Department of Health-led National Awareness and Early Diagnosis Initiative (NAEDI) project in England which coordinates and provides support to activities and research that promote the earlier diagnosis of cancer. Part of this initiative has led to the development of Key messages for ovarian cancer for health professionals which aim to raise awareness of signs and symptoms of the disease and were published in February 00. Most women are diagnosed with advanced (stage II-IV) disease that is associated with poor survival rates. On the other hand a great majority of women with early stage (stage I) ovarian cancer can be cured. Women with ovarian cancer are often suspected of having gastrointestinal disease such as irritable bowel syndrome and therefore not investigated, with resulting delays to diagnosis. However it is now known that women with ovarian cancer experience some symptoms more frequently, more severely and more persistently than women who do not have the disease. Available at: The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

37 0 Clinical question: What are the symptoms and signs of ovarian cancer? Clinical evidence Evidence about symptoms and signs of ovarian cancer came from case control studies. For practical reasons these studies were retrospective and prone to recall bias. For example if women with ovarian cancer can recall their symptom history better than controls, the predictive value of symptoms would be inflated. A systematic review by Bankhead et al., (00) estimated that % [%CI: % to %] of women experienced symptoms before diagnosis. Evidence from case control studies shows that abdominal pain, abdominal distension, urinary symptoms, abdominal mass and postmenopausal/abnormal bleeding are more likely to be reported by women before a diagnosis of ovarian cancer than in women without ovarian cancer (Table.). Table. Individual symptoms for ovarian cancer Symptom Sensitivity Specificity Positive predictive value* Abdominal pain % to % 0% to % Abdominal bloating % to % % to % Abdominal distension Abdominal mass/swelling Urinary frequency or urgency Abnormal or postmenopausal bleeding Loss of appetite % to % % to % % to % % to 0% % to % % to % % to 00% % to % % to % 0% to % 0.0% to 0.% 0.0% to 0.0% 0.0% to.% 0.% to % 0.0% to 0.% 0.% to 0.% 0.0% to 0.% Negative predictive value*.% to.%.% to.%.% to.%.% to.%.% to.% *Assuming a prior probability of undiagnosed ovarian cancer of 0.0% (Hamilton et al, 00) References Friedman et al., 00; Goff et al., 00; Hamilton et al., 00; Kim et al., 00; Lurie et al., 00; Olson et al., 00; Rossing et al., 00; Vine et al., 00 Bankhead et al., 00; Goff et al., 00; Friedman et al., 00; Hamilton et al., 00 Bankhead et al., 00; Goff et al., 00; Friedman et al., 00; Hamilton et al., 00; Lurie et al., 00 Hamilton et al., 00; Lurie et al., 00 Friedman et al., 00; Hamilton et al., 00; Lurie et al., 00; Olson et al., 00; Rossing et al., 00; Vine et al., 00.% Bankhead et al., 00; Friedman et al., 00; Hamilton et al., 00; Lurie et al., 00; Vine 00.% Bankhead et al., 00; Lurie et al., 00; Olson et al., 00; Friedman et al., 00; Hamilton et al., 00 The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

38 Box. Definitions of terms used in this section Sensitivity is the proportion of women with ovarian cancer who experienced the symptom in the year prior to diagnosis. Specificity is the proportion of women without ovarian cancer who did not experience the symptom within the last year. The prior probability or pre-test probability is the background risk that a woman has undiagnosed ovarian cancer, regardless of her symptoms. Hamilton et al., (00) estimated the prior probability of undiagnosed ovarian cancer in women presenting to primary care (for symptoms experienced within the previous year) at 0.0%, using UK national incidence data for ovarian cancer. However, as Hamilton et al., (00) point out, not all women will present to primary care in a given year. In Hamilton s study, 0.% of the control group had not consulted in primary care over the one year period of the study. For women consulting in primary care the prior probability of ovarian cancer was estimated at 0.0%. The positive predictive value (PPV) of a given symptom for ovarian cancer is the proportion of women with that symptom who have ovarian cancer. For example if a symptom had a PPV of 0.% for ovarian cancer, in 00 women with that symptom would have ovarian cancer. The PPV of a symptom for ovarian cancer in those presenting to primary care depends both on the sensitivity/specificity of the symptom and the background risk of ovarian cancer in this population. The negative predictive value (NPV) of a given symptom for ovarian cancer is the proportion of women without that symptom who do not have ovarian cancer. The positive predictive value of bloating as a symptom of ovarian cancer showed great variability, probably due to various definitions of bloating used in the studies (from intermittent temporary bloating to permanent or continued abdominal distension). While the sensitivity of individual symptoms for ovarian cancer is low (see Table.) it can be improved by combining the symptoms (Table.). Hamilton et al., (00) and Rossing et al., (00) noted that % of women with ovarian cancer reported at least one symptom during the year before diagnosis. The Goff symptom index (Goff et al., 00) uses a more restrictive definition of symptoms which incorporates symptom frequency and onset. This improves specificity at the expense of sensitivity. Table. Combining symptoms to improve sensitivity Symptom Sensitivity Specificity Positive predictive value* Any % % to symptom % Goff symptom index % to % % to % 0.% to 0.% 0.0% to 0.% Negative predictive value* More than.% References Hamilton et al., (00); Rossing et al., (00).% Rossing et al., (00); Goff et al., (00); Andersen et al., (00); Kim et al., (00) * Assuming a prior probability of undiagnosed ovarian cancer of 0.0% (Hamilton et al., 00). Any of the following symptoms for at least a week during the previous year: urinary frequency/urgency, abdominal distension, abdominal bloating, pelvic/abdominal pain or loss of appetite. Hamilton et al., (00) also included postmenopausal or rectal bleeding. Rossing et al., (00) also included nausea and diarrhoea/constipation. Any of the following symptoms at least times a month (but present for less than one year): pelvic/abdominal pain, urinary urgency/frequency, increased abdominal size/bloating, and difficulty eating/feeling full (Goff et al., 00). The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

39 Recommendations Refer the woman urgently if physical examination identifies a pelvic or abdominal mass and/or ascites. Carry out tests in primary care (see section. on page ) if a woman (especially if 0 or over) reports having any of the following symptoms on a persistent or frequent basis particularly more than times per month: o persistent abdominal distension (women often refer to this as bloating ) o difficulty eating and/or feeling full (early satiety) o pelvic or abdominal pain o increased urinary urgency and/or frequency. Consider carrying out tests in primary care (see section. on page ) if a woman reports having abnormal vaginal bleeding, unexplained weight loss, abdominal distension, fatigue or changes in bowel habit. Advise any woman who is not suspected of having ovarian cancer to return if her symptoms become more frequent and/or persistent. Carry out appropriate assessments for ovarian cancer (see section. on page ) in any woman of 0 or over who has symptoms that suggest irritable bowel syndrome (IBS) because IBS rarely presents for the first time in women of this age. Linking evidence to recommendations The GDG placed a high value on obtaining a definitive diagnosis of ovarian cancer. It considered increasing patient and primary care awareness of the symptoms of ovarian cancer to be important. The GDG was aware of the need to achieve a balance between the increased numbers of women undergoing investigation to achieve this and the impact on patient morbidity and finite healthcare resources. The GDG considered that there was reasonable quality, retrospective evidence that certain symptoms and signs, when experienced frequently and persistently, are suggestive of a woman having ovarian cancer. It was agreed that identifying those symptoms and signs which should prompt healthcare professionals to consider ovarian cancer, could lead to earlier diagnosis. The GDG believed that the potential benefits of earlier diagnosis could outweigh the potentially increased demand for investigation of women, and associated anxiety. The GDG noted that none of the existing scoring systems for symptoms were sufficiently accurate on their own to initiate an immediate urgent referral. Therefore the GDG took elements of these scoring systems to identify which symptoms warrant further investigation in primary care. In the absence of comparative analysis data of cost and outcomes. health economic evaluation was not feasible. Duration of symptoms and the effect on stage at presentation It has been suggested that earlier diagnosis in a number of cancers could improve survival outcome (Thomson and Forman, 00). However, the natural history of ovarian cancer is unknown. An urgent referral means that the woman is seen by a specialist member of the multidisciplinary team within the national target in England and Wales for referral for suspected cancer, which is currently weeks. National Institute for Health and Clinical Excellence (00) Irritable bowel syndrome in adults: diagnosis and management of irritable bowel syndrome in primary care. NICE clinical guideline. London: National Institute for Health and Clinical Excellence. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

40 Ovarian cancer is the fifth most common cancer in women. A GP with an average size practice may only see one case of ovarian cancer every five years which makes recognition of the symptoms and early diagnosis more difficult. This may mean that women visit their GPs with symptoms of ovarian cancer on several occasions before these are recognised as significant. It is not known if earlier recognition and referral will translate into earlier stage at diagnosis. However, there is general agreement that early symptom identification, with a high index of suspicion for ovarian cancer, has the potential to improve prognosis. The GDG explored the evidence to assess the relationship between the duration of symptoms prior to diagnosis and the survival rates in ovarian cancer. Clinical question: What is the relationship between the duration of pre-diagnostic symptoms of ovarian cancer and survival? Clinical evidence Duration of symptoms and stage at diagnosis Low quality evidence, from retrospective observational studies, suggests women presenting with advanced ovarian cancer experience a similar duration of symptoms to those presenting with early stage disease. Six studies compared the duration of symptoms according to disease stage at diagnosis (Fruchter et al., ; Menczer et al., 00; Goff et al., 000; Olsen et al., 00; Robinson et al., ; Webb et al., 00). None of these studies found a statistically significant difference between the duration of symptoms of women presenting with early and advanced disease. Olson et al., (00) found the duration of symptoms before diagnosis was shorter in women with advanced stage (III to IV) than for early stage (I to II) ovarian cancer for all their symptom categories, except constipation. This difference was not statistically significant, however, except for diarrhoea. Goff et al., (000) reported that women with early stage disease at diagnosis were less likely to report ignoring their symptoms than women with advanced stage disease at diagnosis (% versus %, P=0.00), although there was no significant difference in the time from symptom onset to diagnosis in early versus advanced stage in their study (P=0.). Neal et al., (00) analysed the stage at diagnosis of patients with ovarian cancer according to their referral pathway. There was no significant difference between the stage at diagnosis of urgent guideline referrals and patients diagnosed through other routes (P=0.). Duration of symptoms, quality of life and survival Notwithstanding the particular importance of this clinical question to patients and healthcare professionals, there was insufficient evidence to say whether the duration of symptoms before diagnosis affects overall survival, quality of life or disease specific survival. Research recommendation Further research should be undertaken on the relationship between the duration and frequency of symptoms in women with ovarian cancer before diagnosis, the stage of disease at diagnosis and subsequent survival. Linking evidence to recommendations The GDG acknowledged the lack of available evidence on the outcomes of interest. However, the GDG placed a high value on the potential benefits to be derived from an The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 0 of

41 improved understanding of the relationship between the duration of symptoms and subsequent outcomes. Examination of all the evidence found no association, one way or the other, between the duration of symptoms on the outcomes studied. However, the GDG felt strongly that this lack of evidence should not preclude timely and appropriate referral. As this clinical question addressed an epidemiological issue it was felt unlikely to lend itself to health economic evaluation.. Asking the right question - first tests The majority of women with symptoms suggestive of ovarian cancer will not have ovarian cancer, so symptoms alone are not sufficient to refer to secondary care. Given the increased emphasis on symptom recognition this has to be combined with effective assessment to enable timely and appropriate referral onto the ovarian cancer pathway. There is considerable variation in practice across the UK as to what tests are currently performed in primary care. In addition many women are referred to other specialists in error. The GDG sought to identify the next steps in primary care, given the resources available to GPs. Further test options included pelvic examination, serum CA or pelvic ultrasound either individually or in combination. Clinical examination is an integral part of the assessment of any woman with symptoms. Whilst this is the case it is also recognised that pelvic examination has limitations in its ability in detecting adnexal pathology. A raised serum CA in younger women is less likely to be related to a diagnosis of ovarian cancer and when elevated in this group, can raise considerable worry for GP and patient alike. A serum CA of >000 IU/ml in an older postmenopausal woman is a highly significant finding that points to some sort of malignancy, the most likely being ovarian or primary peritoneal cancer, although other cancers such as lung or pancreatic cancer cannot be excluded on this one test alone. In addition serum CA levels of several hundred may occur as a consequence of non-malignant conditions such as heart failure. Ultrasound is useful for characterising pelvic disease, however, its unselected use in primary care may place an unsustainable burden on diagnostic resources and is operator dependent. Clinical question: For women with suspected ovarian cancer, what are the most effective first tests in primary care? Clinical evidence There was no direct evidence comparing serum CA, morphological ultrasound and pelvic examination in women with symptoms in primary care. Indirect evidence comes from systematic reviews of these tests in secondary care or in screening studies. Due to the differences in case mix between these settings it is likely that the tests will perform differently in each place. Assuming a prevalence of ovarian cancer in women with symptoms presenting to primary care of 0.%, the positive predictive values of the individual tests were 0.% for serum CA (Myers et al., 00) and.% for morphological ultrasound (Liu et al., 00). This The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

42 0 0 0 means that around in every 00 women referred to secondary care with positive serum CA or ultrasound would have ovarian cancer. Negative predictive values were 0.0% for serum CA (Myers et al., 00) and 0.0% for morphological ultrasound (Liu et al., 00), suggesting around in every,000 women with negative tests would turn out to have ovarian cancer. The evidence suggested pelvic examination is relatively insensitive for the detection of adnexal masses. Myers et al., (00) estimated that only % of adnexal masses would be detected on pelvic examination. In women with palpable masses (assuming an ovarian cancer prevalence of 0.%), pelvic examination had a positive predictive value of.0% for ovarian cancer and a negative predictive value of 0.0% (Myers et al., 00). If there is disagreement between the individual tests, there is value in combining them. Tests can be combined to improve the overall sensitivity at the cost of specificity (by referring women who are positive on any of the tests). Tests can also be combined to improve specificity at the cost of sensitivity (by only referring women who are positive on all the tests). There was no direct evidence about the performance of combined serum CA, ultrasound and pelvic examination in primary care. The accuracy of combined tests was therefore estimated using the values from the meta-analyses of individual tests and assuming conditional independence between tests. Combining tests to improve sensitivity meant a reduced positive predictive value of 0.% to 0.% but an improved negative predictive value of 0.0 to 0.0% (depending on which combination was used). Using figures from Hamilton et al., (00) and Bankhead et al., (00), approximately 0.% of women with symptoms consistent with ovarian cancer in primary care actually have ovarian cancer. If all women with symptoms were referred to secondary care, around in every 00 women referred would turn out to have ovarian cancer. If women were only referred if they had a positive serum CA test or ultrasound scan (Table. below), then in every referred would have ovarian cancer (assuming conditional independence between serum CA and ultrasound). % of women with ovarian cancer and symptoms would not be referred. If women were only referred when both CA test and ultrasound were positive, then in every referred would have ovarian cancer. % of women with ovarian cancer and symptoms would not be referred at initial presentation. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

43 Table. Distribution of cases according to test results in a theoretical cohort of 00,000 women with symptoms consistent with ovarian cancer presenting to primary care. Assumed prevalence of undiagnosed ovarian cancer is 0.% in women with such symptoms. Ovarian cancer Referral strategy Test result Yes No Proportion with ovarian cancer Refer if CA is positive CA positive, 0.% Don t refer if CA is negative CA negative, 0.0% Refer if ultrasound is positive ultrasound positive,.% Don t refer if ultrasound is negative ultrasound negative,0 0.0% Refer if CA or ultrasound is positive Don t refer if CA and ultrasound are negative Refer if CA and ultrasound are positive Don t refer if CA or ultrasound is negative * assuming conditional independence CA or ultrasound positive* CA and ultrasound negative * CA and ultrasound positive * CA or ultrasound negative*,0 0.%,0 0.0%,.%, 0.0% Health economic evaluation (see Appendix ) This clinical question was highlighted as a priority for economic analysis because of the large number of patients with symptoms suggestive of ovarian cancer. In addition there are significant differences in costs and health outcomes associated with the diagnostic pathway as well as the considerable economic burden of treating ovarian cancer. Economic evaluations of a diagnostic investigation require evidence on a number of issues, including disease prevalence and test accuracy. Furthermore, the accurate estimation of cost-effectiveness of one diagnostic strategy over another requires consideration of downstream treatment effects, health-related preferences (utilities), healthcare resource use and costs. High quality evidence on all relevant parameters is essential, but not always available. When published evidence is sparse, expert opinion can be used to estimate relevant parameters. To test the robustness of the results of the cost-effectiveness analysis, a sensitivity analysis is undertaken. A decision tree was constructed outlining seven strategies of interest: three of the strategies consisted of a single test (pelvic examination, ultrasound and serum CA) and the remaining four strategies were comprised of a combination of tests (pelvic examination + serum CA; pelvic examination + ultrasound; serum CA + ultrasound and pelvic examination + serum CA + ultrasound). A Markov process was embedded in the decision tree to model the recurrence of disease and survival based on the results of the diagnostic tests and the subsequent management of women presenting with symptom(s) of ovarian cancer in a primary care setting. The clinical evidence required to populate the model was obtained from a number of different sources. Prevalence of the disease in primary care was assumed to comprise of linear summation of the prevalence of ovarian and colorectal malignancies and benign gynaecological problems. The estimates of the prevalence of ovarian and colorectal malignancies were obtained from published literature (CancerResearchUK, 00; Hamilton et al., 00). The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

44 0 0 0 The accuracy of the diagnostic procedures, in terms of the corresponding sensitivity and specificity values, were obtained from the systematic reviews of the clinical evidence conducted for this guideline (see clinical evidence in sections. and.) (Hunink and Glasziou 00; Bell et al., ). There was no consistent reporting of the proportion of patients in each treatment arm, as defined by the model structure, in the published literature. Therefore, the estimates of proportion were elicited from the GDG. Effectiveness of treatment in terms of survival and morbidity rates were obtained from published literature (Kosary ; Chien et al., 00; Gerestein et al., 00; Loft et al., ; Venesmaa and Ylikorkala ; International Collaborative Ovarian Neoplasm Group 00). In addition, healthcare resource use associated with providing supportive care and follow-up monitoring were also obtained via GDG consensus. Utility weights were required to estimate quality adjusted life years (QALYs). Estimates of health state utilities specific to ovarian cancer patients were obtained from published studies (Swart et al., 00; Tappenden et al., 00; Drummond et al., 00). The costs considered in the analysis were those relevant to the UK NHS, and included costs of diagnostic investigations (both in primary and secondary care); costs of therapy (surgery, drug acquisition costs and administration costs) and costs associated with healthcare resource use for provision of supportive care and follow-up monitoring. Unit costs were based on NHS Reference Costs 00-0 or the Unit Costs of Health and Social Care (PSSRU, 00). Within health economic evaluation, discounting of costs and health outcomes is standard practice where costs and benefits that accrue in the future are given less weight to those which occur in the present. Following methodological guidance published by NICE, all costs and health outcomes are discounted at.% per year (PSSRU, 00). A summary of expected cost and expected effectiveness estimates associated with each diagnostic strategy in the model is presented in Table.. The cost of the strategies varies widely, ranging from the least expensive strategy (serum CA) at just over,00 to the most expensive (combination of pelvic examination plus serum CA plus ultrasound) at,0 per patient. Health outcomes, measured in terms of QALYs, ranged from 0. for the serum CA strategy to. for the pelvic examination plus serum CA plus ultrasound combination strategy. Serum CA (single test) strategy on average generates 0. QALYs and ultrasound (single test) generates 0. a difference of 0.00 QALYs is an equivalent (on average) of an additional. days of perfect health. Table. Base case total expected cost and QALYs Strategy Cost ( ) Effectiveness (QALY) Serum CA,. 0. ICER Ultrasound,0. 0. (Dominated) Pelvic examination + serum CA, (Dominated) Pelvic examination + ultrasound,. 0. (Dominated) Pelvic examination,. 0. (Dominated) Serum CA + ultrasound,0.. (Dominated) 0 Pelvic examination + ultrasound + serum CA ICER incremental cost-effectiveness ratio,0.. (Dominated) The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

45 All strategies in this analysis are dominated by the serum CA strategy. A strategy is said to be dominated if it is both more costly and less effective than its comparator. A series of one-way sensitivity analyses were conducted to assess the robustness of the study results. One-way sensitivity analysis describes the process of changing one parameter in the model and re-running the model to see how a change in this parameter influences overall results. Five scenarios were considered and are detailed below: nationally-agreed drug discounts a decrease in prevalence of ovarian malignancy in primary care the prevalence of benign gynaecological problem varied over an agreed range (0% - 0%) a decrease in the proportion of patients who are not fit for further treatment following diagnostic investigation an increase in age at the start of the model. The results of the base case analysis were not sensitive to any of the five scenarios outlined above. The effect of applying nationally agreed price discounts did alter the overall expected costs but did not alter the ranking of the most cost-effective strategy. Specifying the parameters as distributions and performing a probabilistic sensitivity analysis showed that the CA strategy did little to alter this conclusion. Similarly, the results of the one-way sensitivity analysis in the other scenarios showed changes in the overall expected costs and health benefits but did not alter the ranking of the cost-effective diagnostic strategy. Recommendations Measure serum CA in primary care in women with symptoms that suggest ovarian cancer (see section. on page ). If serum CA is greater than IU/ml, arrange an ultrasound scan of the abdomen and pelvis. If the ultrasound suggests ovarian cancer, refer the woman urgently 0 for further investigation. Advise any woman who has normal serum CA, or CA greater than IU/ml but a normal ultrasound, to return to her GP for re-assessment if her symptoms persist. Linking evidence to recommendations The recommendations were based on evidence of test performance and a health economic evaluation of the most cost-effective first test. The GDG recognised the need for an initial test using an objective and standardised assessment in symptomatic women because this would reduce observer variability. Serum tumour markers fulfil these criteria. High value was placed on serum CA as it is currently the most widely used and reliable serum tumour marker for ovarian cancer. The GDG acknowledged that the clinical evidence was of limited applicability because it did not come from symptomatic women in primary care. Although this evidence was based on data in a secondary care setting the GDG felt that it was appropriate to apply its use in the primary care setting. The health economic modelling corroborated this view by conducting sensitivity analyses including the effect of changing prevalence. 0 An urgent referral means that the woman is seen by a specialist member of the multidisciplinary team within the national target in England and Wales for referral for suspected cancer, which is currently weeks. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

46 The clinical evidence demonstrated that no single test on its own adequately selected a manageable number of women for referral to secondary care. The combination of raised serum CA and sequential ultrasound of the abdomen and pelvis reduced significantly the number of women who would be referred, though a greater proportion of symptomatic women would be directed to the right pathway in a more timely fashion. Although the trade off in adopting a sequential strategy as recommended means that some women with ovarian cancer would be missed in the first instance, the view of the GDG was that this was a sensible and pragmatic decision as those women whose symptoms persist would subsequently re-attend and be referred. Having identified a sequential testing strategy on clinical evidence, the health economic modelling unequivocally identified that serum CA was the most cost-effective first test as opposed to ultrasound or ultrasound and serum CA in combination. It was recognised that there would be an impact on health service resources and women tested due to the low prevalence of ovarian cancer in the symptomatic patient group. Equally, it was felt that in order to ensure symptomatic women were placed along the correct pathway as soon as possible it could only be achieved using such a sequential testing strategy. References Andersen MR (00). Use of a Symptom Index, CA, and HE to predict ovarian cancer. Gynecol.Oncol. : -. Bankhead CR, Kehoe ST and Austoker J (00). Symptoms associated with diagnosis of ovarian cancer: a systematic review. BJOG. : -. Bankhead CR, Collins C, Stokes-Lampard H, Rose P, Wilson S, Clements A, Mant D, Kehoe ST and Austoker J (00). Identifying symptoms of ovarian cancer: a qualitative and quantitative study. BJOG. : Bell R., Petticrew M., Luengo S., Sheldon TA. () Screening for ovarian cancer: a systematic review. Health Technology Assessment,. (): CancerResearchUK (00) Cancer Stats: Incidence [cited; Available from: www. cancerresearchuk.org Chien P., Khan K. and Mol BW. (00) How to interpret the findings of the evaluate study. BJOG: An International Journal Of Obstetrics And Gynaecology. (): -. Drummond M F. Sculpher MJ., Torrance GW., O'Brien BJ. and Stoddart GL. (00). Methods for the economic evaluation of health care programmes. Oxford: Oxford University Press, England. Friedman GD, Skilling JS, Udaltsova NV and Smith LH (00). Early symptoms of ovarian cancer: a case-control study without recall bias. Fam.Pract. : -. Fruchter RG and Boyce J. () Delays in diagnosis and stage of disease in gynecologic cancer. Cancer Detection & Prevention (-): - Gerestein CG., Damhuis RA., Burger CW. and Kooi GS. (00) Postoperative mortality after primary cytoreductive surgery for advanced stage epithelial ovarian cancer: A systematic review. Gynecologic Oncology. (): -. Goff BA, Mandel L, Muntz HG and Melancon CH. (000) Ovarian carcinoma diagnosis. Cancer (0): 0- Goff BA, Mandel LS, Melancon CH and Muntz HG (00). Frequency of symptoms of ovarian cancer in women presenting to primary care clinics. JAMA. : 0-. Goff BA, Mandel LS, Drescher CW, Urban N, Gough S, Schurman KM, Patras J, Mahony BS and Andersen MR (00). Development of an ovarian cancer symptom index: possibilities for earlier detection.[see comment]. Cancer. 0: -. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

47 Hamilton W, Peters TJ, Bankhead C and Sharp D (00). Risk of ovarian cancer in women with symptoms in primary care: population based case-control study. BMJ. : b- Hunink M. and Glasziou P. (00) Decision making in health and medicine. Cambridge University Press: Cambridge, UK. International Collaborative Ovarian Neoplasm Group (00) Paclitaxel plus carboplatin versus standard chemotherapy with either single-agent carboplatin or cyclophosphamide, doxorubicin, and cisplatin in women with ovarian cancer: the ICON randomised trial. Lancet 0(): 0. Kim MK, Kim K, Kim SM, Kim JW, Park NH, Song YS and Kang SB (00). A hospital-based casecontrol study of identifying ovarian cancer using symptom index. J Gynecol Oncol. 0: -. Kosary CL. () FIGO stage, histology, histologic grade, age and race as prognostic factors in determining survival for cancers of the female gynecological system: an analysis of - SEER cases of cancers of the endometrium, cervix, ovary, vulva, and vagina. Seminars In Surgical Oncology. 0(): -. Liu JZ, Xu YF and Wang JC.(00) Ultrasonography, computed tomography and magnetic resonance imaging for diagnosis of ovarian carcinoma. Eur Journal Radiol (): - Loft A., Andersen TF., Brønnum-Hansen H,, Roepstorff C. and Madsen M. () Early post operative mortality following hysterectomy. A Danish population based study -. British Journal of Obstetrics and Gynaecology. (): -. Lurie G, Thompson PJ, McDuffie KE, Carney ME and Goodman MT (00). Prediagnostic symptoms of ovarian carcinoma: a case-control study. Gynecol.Oncol. : -. Menczer J, Chetrit A, Sadetzki S and the National Israel Ovarian Cancer Group. (00) The effect of symptom duration in epithelial ovarian cancer on prognostic factors. Arch Gynecol & Obstet (): -0 Myers ER, Bastian LA, Havrilesky LJ, Kulasingam SL and Terplan MS, Cline KE, et al. (00) Management of adnexal mass. Evidence Report/Technology Assessment (0): - National Institute for Health and Clinical Excellence (00) Referral guidelines for suspected cancer. NICE clinical guideline. London: National institute for Health and Clinical Excellence. Neal RD, Allgar VL, Ali N, Leese B, Heywood P, Proctor G, et al. (00) Stage, survival and delays in lung, colorectal, prostate and ovarian cancer: comparison between diagnostic routes. Br J Gen Practice (): - Olsen CM, Cnossen J, Green AC and Webb PM. (00) Comparison of symptoms and presentation of women with benign, low malignant potential and invasive ovarian tumors. Eur J Gynaecol Oncol (): -0 Olson SH, Mignone L, Nakraseive C, Caputo TA, Barakat RR and Harlap S. (00) Symptoms of ovarian cancer. Obstet & Gynecol (): - PSSRU (00) Unit Costs of Health and Social Care Robinson E, Mohilever J, Zidan J and Sapir D. () Delay in diagnosis of cancer. Possible effects on the stage of disease and survival. Cancer (000-X (Print), 000-X (Linking), ): -0 Rossing MA, Wicklund KG, Cushing-Haugen KL and Weiss NS (00). Predictive value of symptoms for early detection of ovarian cancer. J.Natl.Cancer Inst. 0: -. Swart AC. et al., on behalf of ICON collaborators. (00) Long-term follow-up of women enrolled in a randomized trial of adjuvant chemotherapy for early stage ovarian cancer (ICON). Journal of Clinical Oncology (Meeting Abstracts). (_suppl): 0 Tappenden P., Chilcott J., Eggington S., Patnick J., Sakai H., and Karnon J. (00) Option appraisal of population-based colorectal cancer screening programmes in England. Gut (): -. Thomson CS, Forman D (00) Cancer survival in England and the influence of early diagnosis: what can we learn from recent EUROCARE results? British Journal of Cancer, 0: S0 S0. Venesmaa,P. and Ylikorkala O. () Morbidity and mortality associated with primary and repeat operations for ovarian cancer. Obstetrics And Gynecology. ():-. Vine MF., Ness RB., Calingaert B., Schildkraut JM and Berchuck A (00) Types and duration of symptoms prior to diagnosis of invasive or borderline ovarian tumor. Gynecol Oncol : -. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

48 Webb PM. (00) Symptoms and diagnosis of borderline, early and advanced epithelial ovarian cancer. Gynecol Oncol (): - The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

49 Establishing the diagnosis in secondary care The objectives of this chapter were:. to estimate the sensitivity, specificity and positive/negative predictive values of serum tumour markers (other than serum CA) in women with suspected ovarian cancer. to determine which malignancy index is the more accurate in assessing the probability of malignant pathology in women with suspected ovarian cancer. to determine which imaging tests should be done in women with suspected ovarian cancer. to determine when it is appropriate for women with suspected advanced ovarian cancer not to have a tissue diagnosis before starting chemotherapy. to determine whether samples from image-guided biopsy or laparoscopic biopsy are the best method of tissue diagnosis before chemotherapy.. Tumour markers: which to use? Tumour markers are a group of proteins, hormones, enzymes, receptors, and other cellular products that are over-expressed by malignant cells. The evidence supporting the use of serum CA as a useful predictive tumour marker in suspected ovarian cancer is strong (see clinical evidence in section.). It is raised in 0% of such women but can also be significantly elevated in other benign and malignant conditions. This review of clinical evidence sought to look at individual tumour markers in addition to serum CA, especially ones which had been developed more recently, to see if any of these might facilitate the diagnosis in women with suspected ovarian cancer, if routinely carried out. These included CEA, CDX, CA -, CA -, AFP, beta-hcg and HE. Clinical question: For women with suspected ovarian cancer, what serum tumour marker tests should be routinely carried out to aid in diagnosis? Clinical evidence The evidence review considered the diagnostic accuracy of the following serum tumour markers CEA, CDX, CA -, CA -, AFP, beta-hcg and HE in comparison to serum CA in women with suspected ovarian cancer. The evidence came from studies of women who had surgery for pelvic tumours with histopathology to confirm their diagnosis. This means that the evidence is not directly applicable to women with symptoms of ovarian cancer in primary care. The overall methodological quality of these studies was moderate to low - most were case series and not designed as prospective diagnostic studies. The reference standard diagnosis (histopathology) was consistently applied but the timing of the serum tumour marker tests and the use of blinding in the interpretation of tests were rarely reported. HE There was consistent evidence, from five studies comparing HE and serum CA in women with pelvic masses, that HE is more sensitive and specific than serum CA for the diagnosis of ovarian cancer (Abdel-Azeez et al., 00; Huhtinen et al., 00; Moore et al., 00; Nolen et al., 00; Shah et al., 00). These five studies included a total of women with ovarian cancer and with benign disease. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

50 Summary ROC curves suggested peak sensitivity/specificity of % for serum CA compared with % for HE. From these figures, for every,000 women referred for diagnosis of a pelvic tumour, using HE instead of serum CA would identify an additional seven patients with cancer with fewer false positives (assuming a 0% prevalence of undiagnosed ovarian cancer in this population (Myers et al., 00)). Five studies looked at the combination of HE and serum CA (Abdel-Azeez et al., 00; Huhtinen et al., 00; Moore et al., 00; Moore et al., 00; Nolen et al., 00). The evidence suggests that the combination of HE and serum CA is more specific, but less sensitive than either marker in isolation. CA. Ten studies, including women with ovarian cancer and,00 with benign disease, compared CA. to serum CA. The pooled results suggested CA. and serum CA have similar peak sensitivity/specificity, % and % respectively. It is clear from the ROC curves, however, that (at least at the diagnostic thresholds used in the studies) CA. has a lower sensitivity and higher specificity than serum CA. Evidence from a further six studies suggests that combining the two markers could increase their specificity, but at the cost of sensitivity. CA. Eight studies including women with malignant tumours and, with benign disease, compared the diagnostic accuracy of CA - and serum CA in women with pelvic masses.the summary ROC curve suggests CA. has relatively low sensitivity for the diagnosis of ovarian cancer, at the diagnostic thresholds used in the studies. CEA, CDX, AFP and beta-hcg Eight studies including, women, reported the diagnostic accuracy of CEA for the diagnosis of ovarian cancer in women with suspected ovarian cancer. Serum CEA was raised in approximately % of women with ovarian cancer (sensitivity %), but specificity varied widely between studies. The literature searches found no studies about the use of the marker CDX. There was a single study each about the use of serum beta-hcg and serum AFP in the diagnosis of ovarian cancer, suggesting low sensitivity for these markers. AFP and hcg are important markers for triage. However, when there is a suspicion of germ cell tumour, particularly in women younger than 0 years or where scan features suggest a germ cell tumour (for example Sturgeon et al., 00). Multiple tumour marker panels Three of the studies (Nolen et al., 00; Moore et al., 00; Abel-Azeez et al., 00) investigated panels combining three or more serum tumour markers. There was no evidence to suggest that multiple tumour markers were much better than the two marker combination of serum CA and HE. Recommendations Measure serum CA in secondary care in all women with suspected ovarian cancer, if this has not already been done in primary care. In women under 0 with suspected ovarian cancer, measure levels of alpha fetoprotein (AFP) and beta human chorionic gonadotrophin (beta-hcg) as well as serum CA, to help identify women with germ cell tumours. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 0 of

51 Linking evidence to recommendations The GDG placed a high value on the outcomes of sensitivity and specificity of the different tumour marker tests for facilitating a diagnosis of ovarian cancer. At this time there is ample evidence supporting the clinical utility of serum CA in diagnosing ovarian cancer. The GDG acknowledged that the methodological quality of the evidence was low, with most studies being case series and not designed as prospective diagnostic or prognostic studies. The GDG noted that although the preliminary data on HE showed it to have a relatively high sensitivity and specificity, it was not in routine clinical use and studies about its diagnostic performance had only recently been published. The GDG therefore did not feel the data on HE was substantial enough to enable it to be recommended instead of serum CA the only serum tumour marker with widely accepted clinical utility in women with ovarian cancer. They therefore recommended the routine use of serum CA. This clinical question was agreed as a medium priority for health economic evaluation because although there are potentially significant cost differences between the different combinations of serum tumour markers used, other clinical questions were considered higher priority for investigation.. Malignancy indices In women with an adnexal mass it is important to distinguish between benign and malignant pathology before surgical treatment. Improving outcomes in gynaecological cancers guidance (Department of Health, ) recommends that women with ovarian cancer be discussed at a multidisciplinary team meeting and be offered, where appropriate, a laparotomy, a full staging procedure and optimal debulking in a cancer centre by a trained gynaecological oncologist. In contrast, women with low or moderate risk of ovarian cancer can be managed by gynaecological cancer leads in a cancer unit. At present, several parameters are available to help distinguish between benign and malignant masses. These include menopausal status or age, ultrasound characteristics with or without Doppler flow assessment and tumour markers such as serum CA. These parameters can be combined to provide risk of malignancy indices that can help to predict the probability of malignancy. At present, none of the currently available tests can provide 00% sensitivity or specificity; however, most of the available prediction models are useful in the pre-operative assessment of the adnexal mass. Clinical question: For women with suspected ovarian cancer, which malignancy index is the most effective? Clinical evidence The evidence for this topic comprised one good quality systematic review of diagnostic studies (Geomini et al., 00) in which the reviewers appraised 0 studies of eighty-three validated risk of malignancy models. By pooling data appropriately the authors concluded that the RMI I proposed by Jacobs et al., (0) was superior in terms of sensitivity and specificity to the other comparators. With a cut-off score of 00, sensitivity = % [%CI: -%] and specificity = % [%CI: -%] and with a cut-off score of 0, sensitivity = % [%CI: -%] and specificity = % [%CI: -0%]. Raza et al., (00) published a rapid communication reporting the results of a prospective observational study that had been conducted in a UK hospital. Using Jacob s RMI I, as modified by Tingulstad et al., () they referred all women with a suspicious mass and a score of 0 directly to the cancer clinic. All patients were first discussed at a MDT meeting and those with a lower RMI score may still have been referred if there were clinical indications of malignancy. Of 0 women in the study were directly referred, of which one The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

52 had benign disease. One woman with a low RMI was referred to the clinic on the basis of having had a suspicious CT scan. With a cut-off score in this very limited population, the RMI I index had sensitivity =.% [%CI: 0.-%] and specificity.% [%CI:.- 00%]. Recommendation Calculate a risk of malignancy index I (RMI I) score (after performing an ultrasound; see section. on page ) and refer all women with an RMI I score of 00 or greater to a specialist multidisciplinary team. Box. Risk of malignancy index RMI I RMI I combines three pre-surgical features: serum CA (CA), menopausal status (M) and ultrasound score (U). The RMI is a product of the ultrasound scan score, the menopausal status and the serum CA level (IU/ml). RMI = U x M x CA The ultrasound result is scored point for each of the following characteristics: multilocular cysts, solid areas, metastases, ascites and bilateral lesions. U=0 (for an ultrasound score of 0), U= (for an ultrasound score of ), U= (for an ultrasound score of -). The menopausal status is scored as = pre-menopausal and = post-menopausal The classification of post-menopausal is women who have had no period for more than one year or women over the age of 0 who have had a hysterectomy. Serum CA is measured in IU/ml and can vary between 0 to hundreds or even thousands of units. Linking evidence to recommendations The GDG noted that there was high-quality evidence that RMI I was the most useful index at identifying women with ovarian cancer compared to other malignancy indices, but only in the secondary care setting. However the GDG recognised that although the evidence showed RMI I to be the more useful index, it did not indicate the optimum cut-off score to use for guiding management. The GDG felt that an RMI I cut-off of 00 should be used because this would ensure access to specialist centres whilst not overburdening them with benign disease (and the additional costs associated with this). It was also noted that the value of the cut-off score used, affected the sensitivity of RMI I relative to the specificity. For example, a low cut-off score could mean that some women who did not have ovarian cancer would be wrongly identified as positive and referred for specialist treatment. Conversely, a high cut-off score could mean that some women who did have ovarian cancer would not be identified or referred for specialist treatment. The GDG agreed that this clinical question was not relevant for health economic evaluation because it is unlikely that the different malignancy indices would have a direct impact on patient outcomes. See Box. for details of how to calculate an RMI I score. Jacobs I, Oram D, Fairbanks J, Turner J, Frost C and Grudzinskas JG (0) A risk of malignancy index incorporating CA, ultrasound and menopausal status for the accurate preoperative diagnosis of ovarian cancer. Br J Obstet Gynaecol, : -. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

53 Research recommendation Further research should be undertaken to determine the optimum RMI I threshold that should be applied in secondary care to guide the management of women with suspected ovarian cancer.. Imaging in the diagnostic pathway: which procedures? Imaging is used to characterise the extent and spread of ovarian cancer. This information can be used for staging and influencing management decisions. In addition it may facilitate image-guided biopsy to enable histological confirmation of diagnosis. Appropriate imaging will also allow a baseline to be established in order that later imaging can assess response to chemotherapy, or assess disease relapse. The principle imaging modalities comprise ultrasound, computerised tomography (CT) scans and magnetic resonance imaging (MRI), all of which have the capacity to characterise adnexal masses and to assess extent of spread and operability. In addition to how well a test functions one should consider other issues such as availability, cost, and safety. Ultrasound has the advantage of being more available, cheaper and safer. Grey-scale ultrasound performs well in identifying simple cystic masses that have a high negative predictive value. It is therefore well placed as an initial test and enables adnexal masses to be triaged into low (not ovarian cancer) and higher risk (suspected ovarian cancer) categories. Women with ovarian cancer can often have associated pleural effusions, which if malignant, have significant staging and possible management implications. CT is the investigation of choice for detection of disease in the thorax. MRI is established as a tool for characterisation of pelvic masses because of its ability to discriminate masses that contain both fat and blood, neither of which are features of malignancy. However, MRI is less available, scan times are much longer, and imaging of the abdomen can be degraded by movement caused by breathing which may affect the sensitivity of detection of omental and peritoneal disease. In higher risk women, further assessment of extent of spread is required to aid management in terms of identifying sites for biopsy and consideration for surgery. A CT scan has the advantage of enabling a more comprehensive assessment of the body, and is superior to MRI and ultrasound for assessment of the sub-diaphragmatic regions, gastro-splenic ligament, lesser sac and retroperitoneal nodal disease; sites of likely spread of ovarian cancer. CT is less operator dependent than ultrasound, and more available than MRI. Finally CT also provides optimal baseline information in order to assess response to chemotherapy and disease relapse. Clinical question: For women with suspected ovarian cancer, what is the most appropriate imaging to be done to determine future management? Clinical evidence Differentiation of benign from malignant ovarian tumours Evidence from good quality diagnostic systematic reviews and meta-analysis (Liu et al 00, Kinkel et al., 000; Kinkel et al., 00; Medeiros et al., 00; Myers et al., 00) suggests the accuracy of combined grey-scale/colour Doppler ultrasound, CT and MRI for the differentiation of benign and malignant ovarian masses, are broadly similar, with sensitivity approaching 0% and specificity exceeding %. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

54 Li et al., (00) note that ultrasound is most accurate in identifying simple cystic masses, and the ultrasound studies in their meta-analysis had a lower prevalence of complex ovarian lesions than the CT and MRI studies. It is possible that the diagnostic utility of MRI and CT is underestimated in the meta-analyses. Kinkel et al., (00) reviewed evidence for imaging in women with indeterminate masses at grey-scale ultrasound, presumably excluding those women with simple cystic masses. In this group of patients MRI had a higher positive predictive value (post-test probability), than CT and combined grey-scale/colour Doppler ultrasound. Staging There was limited evidence about the optimal imaging modality for staging. A prospective multicentre study including 0 women (Tempany et al., 000) concluded that CT and MRI were more accurate than ultrasound for staging ovarian cancer. Prediction of optimal cytoreduction Most of the evidence about the prediction of optimal cytoreduction came from studies using CT (Bristow et al., 000; Byrom et al., 00 Dowdy et al., 00; Ferrandina et al., 00; Forstner et al., ; Gemer et al., 00; Meyer et al., ; Nelson et al., ; Kebapci et al., 00; Jung et al., 00; Qayyum et al., 00) with only one ultrasound study (Testa et al., 00) and two MRI studies (Forstner et al., ; Qayyum et al., 00). Five studies (Nelson et al., ; Bristow et al., 000; Dowdy et al., 00; Quayyum et al., 00; Meyer et al., ) reported models to predict suboptimal cytoreduction on the basis of CT features. Although the authors of these models report reasonable sensitivity and specificity for their models, two independent studies (Axtell et al., 00; Gemer et al., 00) did not validate these findings. The low positive predictive values reported by Axtell et al., (00) and Gemer et al., (00) suggest that most patients predicted to have sub-optimal cytoreduction will in fact be optimally cytoreduced at operation. Recommendations Perform an ultrasound of the abdomen and pelvis as the first imaging test in secondary care for women with suspected ovarian cancer, if this has not already been done in primary care. If the ultrasound suggests ovarian cancer, perform a CT scan of the pelvis, abdomen and thorax to establish the extent of disease. Do not use MRI routinely for assessing women with suspected ovarian cancer. Linking evidence to recommendations The GDG placed a high value on the need to establish a diagnosis of ovarian cancer and to determine the extent of disease to inform multidisciplinary team discussions. There was good quality evidence from systematic reviews on which to base the recommendations on diagnosis. The GDG agreed that the sensitivity and specificity of ultrasound and CT for establishing a diagnosis, were shown to be broadly equivalent, but that the evidence did not specify which of these imaging modalities was the most effective. Given that ultrasound and CT had been shown to have equivalent sensitivity and specificity, and that ultrasound is more readily available, less costly and involves no radiation unlike CT, The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

55 the GDG felt it was appropriate to recommend ultrasound as the initial imaging test for women with suspected ovarian cancer. The GDG noted that the evidence for the staging of ovarian cancer was sparse. The GDG recognised that ultrasound is subjective and operator dependent and has limitations in detecting peritoneal disease, whereas multi-slice CT has high spatial resolution and is more sensitive for assessment of omental and peritoneal disease, and abdominal and pelvic lymph nodes. CT is the investigation of choice for staging thoracic disease. For these reasons the GDG chose CT to be the investigation of choice for staging. MRI is less specific for establishing the extent of disease, it is less available and takes longer than CT or ultrasound. For these reasons the GDG were unable to recommend MRI for routine use. This clinical question was considered as a medium priority for health economic evaluation because the population involved was relatively small and the cost difference between the competing alternatives was minimal. Research recommendation Large multicentre case control studies should be conducted to compare the accuracy of CT versus MRI for staging and for predicting optimal cytoreduction.. Tissue diagnosis Requirement for tissue diagnosis Without a tissue diagnosis there is always a degree of diagnostic uncertainty. In most instances, histology is the only way of determining the cancer type and grade and will also exclude other diagnoses such as tuberculosis, inflammation, fibrosis and other infections. Different histological types of ovarian cancer require different treatments, and so confirmed histological diagnosis is considered important. Histological diagnosis is usually made following surgery. In some cases, for example where surgery is not feasible or where chemotherapy is the initial treatment, other options for obtaining a histological diagnosis may be considered. There are a range of methods of obtaining a tissue diagnosis including needle biopsy, laparoscopy or open laparotomy. All are invasive and therefore carry risks. In addition, attempts at tissue diagnosis are not always successful and this may delay the start of treatment. Another method of obtaining a tissue diagnosis is the use of frozen section at the time of surgery. However, this suffers from sampling error and is not widely practised in the UK. Cytology (examination of individual cells aspirated from intra-abdominal fluid or rarely from a tumour) is generally safer than tissue biopsy but has a lower diagnostic accuracy. When it is hazardous or difficult to obtain a tissue diagnosis, the risks of such procedures need to be weighed against the potential benefits of greater diagnostic accuracy. After discussion with the woman it may be concluded that a tissue diagnosis is not essential. Clinical question: For women with suspected advanced ovarian cancer, when is it appropriate not to have a tissue diagnosis before starting chemotherapy? The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

56 Clinical evidence There were no studies comparing the outcomes of women with suspected versus confirmed advanced ovarian cancer treated with chemotherapy. Evidence from case series suggests a minority of women ( %) with presumed advanced ovarian cancer on the basis of clinical and imaging findings will not have ovarian cancer (Griffin et al., 00; Freedman et al., 00). Thus if tissue diagnosis were omitted some women might receive inappropriate treatment. Cytomorphology combined with immunocytochemistry had a rate of definitive diagnosis of primary tumour site in malignant effusions ranging from % to % (Mottolese et al., ; Pomjanski et al., 00; Longatto-Filho et al., ; DiBonito et al,. ). In comparison, histopathology plus immunohistochemistry had a diagnostic rate between % and % in women with peritoneal carcinomatosis of unknown origin (Hewitt et al., 00; Spencer et al., 00) or presumed advanced ovarian cancer (Griffin et al., 00). There were no data about complications of effusion cytology. Percutaneous core biopsy was associated with minor local bruising and discomfort (Fisherova et al., 00; Griffin et al., 00; Hewitt et al., 00; Pombo et al., ; Spencer et al., 00). There was no direct evidence about the harms of diagnostic laparoscopy or laparotomy in women with suspected advanced ovarian cancer due to receive chemotherapy. Indirect evidence comes from studies reporting diagnostic laparoscopy in patients with ascites of unknown origin (Bedioui et al., 00; Chu et al., ; Yoon et al., 00). Minor complications were reported in less than two percent of laparoscopies. Major complications occurred at a rate of less than one percent. Recommendations Obtain a confirmed tissue diagnosis before offering cytotoxic chemotherapy to women with suspected advanced ovarian cancer in all but exceptional cases (see recommendation below). Offer cytotoxic chemotherapy for suspected advanced ovarian cancer without a confirmed tissue diagnosis only: o in exceptional cases, after discussion at the multidisciplinary team o after discussing with the woman the possible benefits and risks of starting chemotherapy without a tissue diagnosis. Linking evidence to recommendations The GDG noted that the evidence for this clinical question consisted of small retrospective studies of moderate quality. The GDG felt that having a histological diagnosis was essential to guiding future treatment, but recognised that on occasions the risks of obtaining a tissue diagnosis might not be justified. In these circumstances, the risk of giving chemotherapy when the diagnosis is uncertain has to be weighed against the potential risks of obtaining histological confirmation. This clinical question was agreed as a low priority for health economic evaluation because of the lack of good quality prospective clinical studies in this area. Methods of tissue diagnosis other than laparotomy Image-guided biopsy is usually performed under local anaesthetic in the radiology department using ultrasound or CT to sample an accessible area of abnormality such as a peritoneal deposit or omental disease. The biopsy needle is inserted percutaneously and The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

57 several passes are usually made to obtain thin tissue cores. This technique is not suitable for all women, for example if the disease is not in an accessible location. It is associated with minor complications, such as local bruising and discomfort. Targeting of the abnormality for biopsy is limited by the imaging technique used and the samples are much smaller, reducing the diagnostic yield. This potentially results in a lower success rate requiring a repeat procedure or surgical biopsy. When image-guided biopsy is not appropriate or if the procedure has failed to obtain an adequate sample, a secondary intervention may be required to obtain tissue for diagnosis. Laparoscopy is a surgical technique that uses an endoscope that gives a complete view but full visualisation of the peritoneal cavity and allows a biopsy to be performed. It requires a general anaesthetic and is more complex to perform. Laparoscopy is associated with both major and minor complications, with higher associated major complication rates than imageguided biopsy. Both techniques have the potential to damage the abdomino-pelvic organs which may be displaced or tethered to abnormal positions by tumour, fibrosis or inflammation. There is also a potential risk of tumour being deposited along the biopsy needle track or implanted into the laparoscopic surgery sites. Clinical question: What is the best method of tissue diagnosis before chemotherapy, samples from image-guided biopsy or laparoscopic biopsy? Clinical evidence The literature search found no studies directly comparing image-guided with laparoscopic biopsy. Evidence from case series indicates a definitive diagnostic rate between % and % for image-guided biopsy (Griffin et al., 00; Hewitt et al., 00; Spencer et al., 00), but our searches found no studies reporting the diagnostic yield of laparoscopic biopsy. Percutaneous core biopsy was associated with minor local bruising and discomfort. Minor complications were reported in less than two percent of laparoscopies from three series (Dedioui et al., 00, Chu et al., ; Yoon et al., 00) with, patients (including cases with non-malignant aetiology). Major complications occurred at a rate of less than one percent. Recommendations Use biopsy rather than cytology to obtain tissue for diagnosis if surgery has not been performed: o use percutaneous image-guided biopsy if this is feasible o use laparoscopy only if percutaneous image-guided biopsy is not feasible or has not produced an adequate sample. Linking evidence to recommendations There was low quality evidence, with no studies directly comparing image-guided biopsy with laparoscopic biopsy, and so case series evidence for the risks and accuracy of each technique in isolation was reviewed. The GDG acknowledged that although there was evidence for the diagnostic yield of imageguided biopsy there was none reporting the diagnostic yield of laparoscopic biopsy. They also noted that higher associated major complication rates were reported with laparoscopic biopsy than image-guided biopsy. The GDG therefore put a high value on the outcomes of morbidity and adverse events associated with the two techniques, and agreed that the simplest and least invasive technique was image-guided biopsy. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

58 This clinical question was originally agreed a high priority for health economic evaluation because the number of patients involved could potentially be large and there could be significant cost implications. Due to the lack of comparative clinical evidence, which would hinder the development of a robust economic analysis it was reconsidered as a low priority. Economic evaluation based on poor quality data would carry a high level of uncertainty and potentially limit its usefulness in informing recommendations. References Abdel-Azeez HA, Labib HA, Sharaf SM and Refai AN (00). HE and mesothelin: novel biomarkers of ovarian carcinoma in patients with pelvic masses. Asian Pacific Journal of Cancer Prevention: Apjcp. : -. Axtell AE, Lee MH, Bristow RE, Dowdy SC, Cliby WA, Raman S, Weaver JP, Gabbay M, Ngo M, Lentz S, Cass I, Li AJ, Karlan BY and Holschneider CH (00) Multi-institutional reciprocal validation study of computed tomography predictors of suboptimal primary cytoreduction in patients with advanced ovarian cancer J.Clin.Oncol. : -. Bedioui, H., Ksantini R.,, N. K., Mekni, A., & Chebbi, F. (00). Role of laparoscopic surgery in the etiologic diagnosis of exudative ascites: a prospective study of 0 cases. Gastroenterol Clin Biol [], - Bristow RE, Duska LR, Lambrou NC, Fishman EK, O'Neill MJ, Trimble EL and Montz FJ (000) A model for predicting surgical outcome in patients with advanced ovarian carcinoma using computed tomography Cancer : -0. Byrom J, Widjaja E, Redman CW, Jones PW and Tebby S (00) Can pre-operative computed tomography predict resectability of ovarian carcinoma at primary laparotomy?[see comment] BJOG: Int J Obstet Gynaecol 0: -. Chu CM, Lin SM, Peng SM, Wu CS, Liaw YF.() The role of laparoscopy in the evaluation of ascites of unknown origin. Gastrointestinal Endoscopy 0(): - Department of Health () Improving outcomes in gynaecological cancers. Service guidance. Available from DiBonito, L., Falconieri, G., Colautti, I., Bonifacio, D., & Dudine, S. (). The positive peritoneal effusion. A retrospective study of cytopathologic diagnoses with autopsy confirmation. Acta Cytologica,, -. Dowdy SC, Mullany SA, Brandt KR, Huppert BJ and Cliby WA (00) The utility of computed tomography scans in predicting suboptimal cytoreductive surgery in women with advanced ovarian carcinoma Cancer 0: -. Ferrandina G, Sallustio G, Fagotti A, Vizzielli G, Paglia A, Cucci E, Margariti A, Aquilani L, Garganese G and Scambia G (00) Role of CT scan-based and clinical evaluation in the preoperative prediction of optimal cytoreduction in advanced ovarian cancer: a prospective trial Br.J.Cancer 0: 0-0. Fischerova D, Cibula D, Dundr P, Zikan M, Calda P, Freitag P and Slama J (00) Ultrasound-guided tru-cut biopsy in the management of advanced abdomino-pelvic tumors International Journal of Gynecological Cancer : -. Forstner R, Hricak H, Occhipinti KA, Powell CB, Frankel SD and Stern JL () Ovarian cancer: staging with CT and MR imaging Radiology : -. Freedman OC, Dodge J, Shaw P, Oza, AM., Bernadini M and Klachook S (00). Diagnosis of epithelial ovarian carcinoma prior to neoadjuvant chemotherapy. Gynecologic Oncology. in press Geomini P, Kruitwagen R, Bremer GL, Cnossen J and Mol BW. (00) The accuracy of risk scores in predicting ovarian malignancy: a systematic review. Obstet Gynecol ( Pt ): - Gemer O, Gdalevich M, Ravid M, Piura B, Rabinovich A, Gasper T, Khashper A, Voldarsky M, Linov L, Ben S, I, Anteby EY and Lavie O (00) A multicenter validation of computerized tomography The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

59 models as predictors of non- optimal primary cytoreduction of advanced epithelial ovarian cancer Eur.J.Surg.Oncol. : 0-. Griffin N, Grant LA, Freeman SJ, Jimenez-Linan M, Berman LH, Earl H, Ahmed A, Crawford R, Brenton J and Sala E (00) Image-guided biopsy in patients with suspected ovarian carcinoma: a safe and effective technique? Eur.Radiol. : 0-. Hewitt MJ, Hall GD, Wilkinson N, Perren TJ, Lane G and Spencer JA (00) Image-guided biopsy in women with breast cancer presenting with peritoneal carcinomatosis International Journal of Gynecological Cancer : 0-0. Huhtinen K, Suvitie P, Hiissa J, Junnila J, Huvila J, Kujari H, et al.(00) Serum HE concentration differentiates malignant ovarian tumours from ovarian endometriotic cysts. Br J Cancer 00(): - Jacobs I., Oram D., Fairbanks J., Turner J., Frost C and Grudzinskas JG (0) A risk of malignancy index incorporating CA, ultrasound and menopausal status for the accurate preoperative diagnosis of ovarian cancer. Br J Obstet Gynaecol : -. Jung DC, Kang S, Kim MJ, Park SY and Kim HB (00). Multidetector CT predictors of incomplete resection in primary cytoreduction of patients with advanced ovarian cancer. Eur.Radiol. 0: Kebapci M, Akca AK, Yalcin OT, Ozalp SS, Calisir C and Mutlu F (00). Prediction of suboptimal cytoreduction of epithelial ovarian carcinoma by preoperative computed tomography. Eur.J.Gynaecol.Oncol. : -. Kinkel K, Hricak H, Lu Y, Tsuda K and Filly RA (000) US characterization of ovarian masses: a metaanalysis. DARE Structured Abstract available Radiology : 0-. Kinkel K, Lu Y, Mehdizade A, Pelte MF and Hricak H (00) Indeterminate ovarian mass at US: incremental value of second imaging test for characterization--meta-analysis and Bayesian analysis Radiology : -. Liu J, Xu Y and Wang J (00) Ultrasonography, computed tomography and magnetic resonance imaging for diagnosis of ovarian carcinoma Eur.J.Radiol. : -. Longatto FA, Bisi H, Alves VA, Kanamura CT, Oyafuso MS, Bortolan J and Lombardo V (). Adenocarcinoma in females detected in serous effusions. Cytomorphologic aspects and immunocytochemical reactivity to cytokeratins and 0. Acta Cytol. : -. Medeiros LR, Rosa DD, da Rosa MI and Bozzetti MC (00) Accuracy of ultrasonography with color Doppler in ovarian tumor: a systematic quantitative review International Journal of Gynecological Cancer : -0. Meyer JI, Kennedy AW, Friedman R, Ayoub A and Zepp RC () Ovarian carcinoma: value of CT in predicting success of debulking surgery AJR : -. Mottolese M, Venturo I, Donnorso RP, Curcio CG, Rinaldi M, Natali PG. () Use of selected combinations of monoclonal antibodies to tumor associated antigens in the diagnosis of neoplastic effusions of unknown origin. Eur J Cancer Clin Oncol. (): - Moore RG, Brown AK, Miller MC, Skates S, Allard WJ, Verch T, et al. (00) The use of multiple novel tumor biomarkers for the detection of ovarian carcinoma in patients with a pelvic mass. Gynecol Oncol 0(): 0- Moore RG, McMeekin DS, Brown AK, DiSilvestro P, Miller MC, Allard WJ, et al. (00) A novel multiple marker bioassay utilizing HE and CA for the prediction of ovarian cancer in patients with a pelvic mass. Gynecol Oncol (): 0- Myers ER, Bastian LA, Havrilesky LJ, Kulasingam SL, Terplan MS, Cline KE, Gray RN and McCrory DC (00). Management of adnexal mass. Evid.Rep.Technol.Assess.(Full.Rep.). -. Nelson BE, Rosenfield AT and Schwartz PE () Preoperative abdominopelvic computed tomographic prediction of optimal cytoreduction in epithelial ovarian carcinoma J.Clin.Oncol. : -. Nolen B, Velikokhatnaya L, Marrangoni A, De GK, Lomakin A, Bast RC, Jr. and Lokshin A (00). Serum biomarker panels for the discrimination of benign from malignant cases in patients with an adnexal mass. Gynecol.Oncol. : 0-. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

60 0 0 0 Pombo F, Rodriguez E, Martin R and Lago M () CT-guided core-needle biopsy in omental pathology Acta Radiol. : -. Pomjanski N, Grote HJ, Doganay P, Schmiemann V, Buckstegge B, Bocking A. (00) Immunocytochemical identification of carcinomas of unknown primary in serous effusions. Diagnostic Cytopathology. (): 0- Qayyum A, Coakley FV, Westphalen AC, Hricak H, Okuno WT and Powell B (00) Role of CT and MR imaging in predicting optimal cytoreduction of newly diagnosed primary epithelial ovarian cancer Gynecol.Oncol. : 0-0. Raza A, Mould T, Wilson M, Burnell M, Bernhardt L. (00) Increasing the effectiveness of referral of ovarian masses from cancer unit to cancer center by using a higher referral value of the risk of malignancy index. Int J Gynecol Cancer 0(): -. Shah CA, Lowe KA, Paley P, Wallace E, Anderson GL, McIntosh MW, et al. (00) Influence of ovarian cancer risk status on the diagnostic performance of the serum biomarkers mesothelin, HE, and CA. Cancer Epidemiology, Biomarkers & Prevention (): - Spencer JA, Swift SE, Wilkinson N, Boon AP, Lane G and Perren TJ (00) Peritoneal carcinomatosis: Image-guided peritoneal core biopsy for tumor type and patient care Radiology : -. Sturgeon CM, Duffy MJ, Stenman UH, Lilja H, Brunner N, Chan DW, et al. (00) National Academy of Clinical Biochemistry laboratory medicine practice guidelines for use of tumor markers in testicular, prostate, colorectal, breast, and ovarian cancers. Clin Chem (): e- Tempany CM, Zou KH, Silverman SG, Brown DL, Kurtz AB and McNeil BJ (000) Staging of advanced ovarian cancer: comparison of imaging modalities--report from the Radiological Diagnostic Oncology Group Radiology : -. Testa AC, Ludovisi M, Savelli L, Fruscella E, Ghi T, Fagotti A, Scambia G and Ferrandina G (00) Ultrasound and color power Doppler in the detection of metastatic omentum: a prospective study Ultrasound in Obstetrics & Gynecology : -0. Tingulstad S., Hagen B., Skjeldestad FE., Onsrud M., Kiserud T., Halvorsen T and Nustad K () Evaluation of a risk of malignancy index based on serum CA, ultrasound findings and menopausal status in the pre-operative diagnosis of pelvic masses. Br J Obstet Gynaecol 0: -. Yoon YJ, Ahn SH, Park JY, Chon CY, Kim do Y, Park YN, et al. (00) What is the role of diagnostic laparoscopy in a gastroenterology unit? J Gastroenterol (): - The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 0 of

61 Management of suspected early (stage I) ovarian cancer The two objectives of this chapter were:. to determine whether removal of the retroperitoneal lymph nodes during standard surgical treatment for suspected early stage ovarian cancer would confer any added benefit to adjuvant therapy. to determine the clinical benefits and toxicity of first-line adjuvant chemotherapy for women with stage I ovarian cancer.. Staging: the role of systematic retroperitoneal lymphadenectomy In women whose disease is thought to be confined to the ovary(s), optimum surgical staging comprises midline laparotomy to allow thorough assessment of the abdomen and pelvis; a total abdominal hysterectomy, bilateral salpingo-oophorectomy and infracolic omentectomy; biopsies of any peritoneal deposits; random biopsies of the pelvic and abdominal peritoneum and retroperitoneal lymph node assessment (Winter-Roach et al., 00). In women where the disease appears to be confined to one ovary and who wish to conserve fertility, then conservative surgery can be considered where the uterus and contra-lateral ovary are conserved. It is recognised that around % of women considered to have stage I ovarian cancer, will in fact have occult retroperitoneal lymph node metastases which can only be identified by removing affected nodes (Maggioni et al., 00). Current surgical staging guidelines advocate only sampling a number of pelvic and/or para-aortic nodes but inevitably less will be sampled than at a systematic retroperitoneal lymphadenectomy, which aims to remove all pelvic and para-aortic lymph nodes up to the renal vessels. Removing all affected nodes will improve staging and might be therapeutic. Systematic retroperitoneal lymphadenectomy is a major surgical procedure which carries the potential risks of prolonged anaesthesia and surgical complications such as increased blood loss and transfusion, ureteric injury, lymphoedema, lymphocysts, damage to nerves and major vessels. There is no international agreement on whether the potential survival benefits of systematic retroperitoneal lymphadenectomy outweigh the risks. Clinical question: For women with ovarian cancer whose disease appears confined to the ovaries, what is the effectiveness of systematic retroperitoneal lymphadenectomy in surgical management? Clinical evidence The evidence for this topic was generally of low quality, comprising two retrospective observational studies, one non-randomised comparative study and a small randomised controlled trial (RCT) (Table.). Across all studies, the majority of women had stage I ovarian cancer. Only the RCT reported the incidence of post-surgical morbidity and none of the papers reported on patient quality of life. The results of survival outcomes were inconsistent between studies. Maggioni et al., (00) presented results from a small, underpowered study that was unable to demonstrate a difference in short or long term survival between patients having surgery The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

62 0 0 alone or surgery with systematic lymphadenectomy (SL). But the more extensive operation was associated with increased morbidity. Conversely, Yokoyama et al., () found a significant difference in the rates of and 0 year survival for women with stage I/II disease who had received SL compared with those who had not (00% vs..% (P<0.0) and.% vs..% (P<0.0) respectively). These results may have been confounded by the addition of different chemotherapy regimens to the study arms. The retrospective studies also reported conflicting results for survival. The largest study (Chan et al., 00; N=,) found a significant improvement in the rate of year diseasespecific survival for women who underwent SL as part of staging compared with women who did not (.% ± 0. vs. % ± 0. P<0.00). However, during the study period participants had unrecorded treatments including surgery and/or chemotherapy which could have confounded these results. The smaller study (Yang et al., 00) found no significant differences in survival after,, or 0 years between women that had undergone SL after primary surgery and those who had not. Again, some participants had subsequently received chemotherapy which could have confounded the results. Kim et al., (00) conducted a thorough systematic review and meta-analysis of RCTs and observational studies to determine the possible benefit of systematic retroperitoneal lymphadenectomy to women with all stages of ovarian cancer. A sub-set of patients had stage I-II disease and these data showed a survival advantage with SL (HR: 0.0 [%CI: ] (P=0.00) with no between studies heterogeneity. However, the included studies were not of high evidential quality consisting of Chan et al., 00; Maggioni et al., 00 and a small retrospective observational study (Suzuki et al., 00). The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

63 Table. GRADE profile: For women with ovarian cancer whose disease appears confined to the ovaries, what is the effectiveness of systematic retroperitoneal lymphadenectomy in surgical management? Quality Summary of findings No of studies Design Limitations Inconsistency Indirectness Imprecision Other Ppts with SL Ppts with no SL % survived SL % survived no SL Quality year disease-specific survival. All study participants (P<0.00) Chan et al. (00). retrospective observational study N/A N/A N/A N/A nil,,. ± 0. ± 0. VERY LOW year disease-specific survival. Age >0 years (P<0.00) Chan et al. (00). retrospective observational study N/A N/A N/A N/A nil,,0 ± 0.. ± 0. VERY LOW year disease-specific survival. Non-clear cell epithelial carcinoma (P<0.00) Chan et al. (00). retrospective observational study N/A N/A N/A N/A nil,,00. ± 0.. ± 0. VERY LOW year disease-specific survival. No hysterectomy (P<0.00) Chan et al. (00). retrospective observational study N/A N/A N/A N/A nil 0,0. ± 0..0 ± 0. VERY LOW year disease-specific survival. Hysterectomy (P=0.0) Chan et al. (00). retrospective observational study N/A N/A N/A N/A nil,,. ± 0.. ± 0. VERY LOW year disease-specific survival. No surgery (P=0.0) Chan et al. (00). retrospective observational study N/A N/A N/A N/A nil 00 ± 0.0. ±. VERY LOW year disease-specific survival. Stage I disease (P<0.00) Chan et al. (00). The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

64 Quality Summary of findings No of studies Design Limitations Inconsistency Indirectness Imprecision Other retrospective observational study Ppts with SL Ppts with no SL % survived SL % survived no SL N/A N/A N/A N/A nil. ±.. ±. Quality VERY LOW year disease-specific survival. Grade disease (P<0.00) Chan et al. (00). retrospective observational study N/A N/A N/A N/A nil. ±.. ±.0 VERY LOW year disease-specific survival. No radiation therapy (P<0.00) Chan et al. (00). retrospective observational study N/A N/A N/A N/A nil,,. ± 0.. ± 0. VERY LOW year disease-specific survival. Caucasian race (P<0.00) Chan et al. (00). retrospective observational study N/A N/A N/A N/A nil,,0. ± 0.. ± 0. VERY LOW year survival stage I (% only) Yang et al. (00) retrospective observational study N/A N/A N/A N/A nil.. VERY LOW year survival stage I (% only) Yang et al. (00) retrospective observational study N/A N/A N/A N/A nil.. VERY LOW year survival stage I (% only) Yang et al. (00) retrospective observational study N/A N/A N/A N/A nil.. VERY LOW 0 year survival stage I (% only) Yang et al. (00) The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

65 Quality Summary of findings No of studies Design Limitations Inconsistency Indirectness Imprecision Other retrospective observational study Ppts with SL Ppts with no SL % survived SL % survived no SL N/A N/A N/A N/A nil..0 Quality VERY LOW year survival stage II (% only) Yang et al. (00) retrospective observational study N/A N/A N/A N/A nil.. VERY LOW year survival stage II (% only) Yang et al. (00) retrospective observational study N/A N/A N/A N/A nil.. VERY LOW year survival stage II (% only) Yang et al. (00) retrospective observational study N/A N/A N/A N/A nil.. VERY LOW 0 year survival stage II (% only) Yang et al. (00) retrospective observational study N/A N/A N/A N/A nil. 0. VERY LOW Estimated year survival for stages I and II (% only) Yokoyama et al. () nonrandomised comparative study N/A N/A N/A N/A nil VERY LOW Estimated 0 year survival for stages I and II (% only) Yokoyama et al. () nonrandomised comparative study N/A N/A N/A N/A nil 0.. VERY LOW The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

66 Quality Summary of findings No of studies Design Limitations Inconsistency Indirectness Imprecision Other SL No SL Relative effect Absolute effect Quality Risk of death. All participants (P>0.0) Maggioni et al. (00) randomised controlled trial limitations N/A N/A underpowered study N/A 0 HR=0. (0.-.) - LOW Risk of progression All participants (P>0.0) Maggioni et al. (00) randomised controlled trial limitations N/A N/A underpowered study N/A 0 HR=0. (0.-.) - LOW year overall survival Maggioni et al. (00) randomised controlled trial limitations N/A N/A underpowered study N/A %.% MD=. (-.-.) LOW year progression-free survival Maggioni et al. (00) randomised controlled trial limitations N/A N/A underpowered study N/A.%.% MD=. (-.-.) - LOW Overall survival. Kim et al., (00) 0 randomised trial and observational studies serious limitations inconsistency indirectness imprecision N/A - - HR=0.0 (0.0-0.) - MODERATE Footnotes: This study combined one small RCT and two observational studies which showed no between studies heterogeneity (0%) and gave a significant result. Nonetheless, the included studies were themselves between low and moderate quality. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

67 Recommendation Do not include systematic retroperitoneal lymphadenectomy as part of the standard surgical treatment of suspected ovarian cancer in women whose disease appears to be confined to the ovaries (that is, who appear to have stage I disease). Linking evidence to recommendations The GDG acknowledged that evidence on the basis of study quality assessed according to GRADE was limited and of poor quality. There was no survival benefit from systematic retroperitoneal lymphadenectomy. They also noted that no studies reported on quality of life. The GDG noted the complications and likely increased costs associated with performing systematic retroperitoneal lymphadenectomy and were unable to recommend its use in women whose disease appears to be confined to the ovaries. This clinical question was agreed as a low priority for health economic evaluation because of the lack of good quality RCT data in this area. Also, given that an economic evaluation would be unlikely to clarify the uncertain health benefits associated with these interventions, the added value of such an analysis was lower than for other clinical questions. Research recommendation A prospective randomised trial should be undertaken to evaluate the cost effectiveness and associated risks of systematic retroperitoneal lymphadenectomy in women with ovarian cancer whose disease appears to be confined to the ovaries.. Adjuvant systemic chemotherapy in stage I disease: patient selection No surgical staging procedure is perfect and in a proportion of women in whom the disease is thought to be confined to the ovaries and completely removed at operation there may, in fact, be occult residual disease. In women with apparent stage I disease, chemotherapy can be given in certain circumstances, such as poorly differentiated tumours and in certain histological sub-types (for example, clear cell carcinomas). This is done to treat residual disease that is suspected but may not, in fact, exist. Therefore some women without residual disease will receive chemotherapy with its associated risks. Given that women with stage I ovarian cancer have significantly less disease it is possible that less chemotherapy will be required for cure. Currently NICE technology appraisal guidance (NICE, 00) recommends a choice of either platinum based compound on its own or in combination with paclitaxel (see section.) but does not stipulate the number of cycles to be given. It is logical that reducing the number of cycles of chemotherapy is likely to reduce toxicity but could compromise effectiveness. The GDG felt that establishing the evidence base for reducing chemotherapy cycles should be investigated in order to quantify any risk-benefit assessment. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

68 Clinical question: For women with stage I ovarian cancer, what is the most effective first line chemotherapy? Clinical evidence The evidence for this topic consisted of one high quality Cochrane review and a lower quality randomised controlled trial (RCT) (Table.). Across these studies, women had undergone primary surgery and had stage I or II ovarian cancer. Winter-Roach et al., (00) conducted a review which investigated whether adjuvant therapy with mainly platinum-containing regimes was associated with a survival advantage compared to withholding chemotherapy until disease progression, and whether certain subgroups of patients gained more or less from this approach. After an average follow-up of nearly ten years it was found that women receiving adjuvant therapy had a considerable advantage in overall survival (HR=0. [%CI: 0. to 0.] P=0.0) and progression-free survival (HR=0. [%CI: 0.-0.] P=0.000). In particular, those women who had been adequately staged gained no survival advantage from immediate adjuvant chemotherapy (HR=. [%CI: 0.-.] P=0.) whereas women who had been inadequately staged did (HR=0. [%CI: 0. to 0.] P=0.00). Bell et al., (00) compared six vs. three cycles of adjuvant carboplatin and paclitaxel in women with early stage ovarian cancer (N=). Across all patients and after an average follow-up of. years, there were no statistically significant differences in the risk of death (HR=.0 [%CI: 0.-.] P=0.) or the rate of disease recurrence (HR=0. [%CI: 0.-.] P=0.). The higher number of treatment cycles was associated with significantly increased morbidity. The systematic review (Winter-Roach et al., 00) included evidence from the Adjuvant Chemotherapy in Ovarian neoplasm (ACTION) trial which has now been updated by Trimbos et al., (00). The results showed that, even with observation, optimally surgically staged patients had a significantly better prognosis compared with patients who had been non-optimally staged: cancer-specific survival (risk of death: HR. [%CI:.-.] (P=0.00); recurrence-free survival (risk of death: HR. [%CI:.-.] P=0.00). In non-optimally staged patients only, adjuvant chemotherapy provided significantly improved cancer-specific survival (risk of death: HR 0. [%CI: 0.-0.] P=0.0) and recurrencefree survival (risk of death: HR 0.0 [%CI: 0.-0.] P=0.00) when compared with observation. The authors concluded, therefore, that the benefit of adjuvant chemotherapy appeared to be limited to patients with non-optimal staging who, perhaps, had a greater risk of unidentified residual disease. The results of Bell et al., 00 were re-analysed in a more recent report (Chan et al., 00) after a median follow-up of months. The authors grouped data by tumour type (i.e. serous or non-serous) and showed that only women with serous cancer derived a significant benefit from six cycles compared with three cycles of adjuvant carboplatin and paclitaxel chemotherapy (HR=0. [%CI: 0.-0.] P=0.00). Although interesting, the original study was underpowered for sub-group analyses which, in any event, have been performed post hoc. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

69 Table. GRADE profile: For women with stage I ovarian cancer, what is the most effective first line chemotherapy No of studies Quality assessment Design Limitations Inconsistency Indirectness Imprecision Other No of patients Chemotherapy Observation Summary of findings Relative (% CI) Effect Absolute Quality OS years. Follow-up -0 months. Winter-Roach et al (00) randomised trials limitations inconsistency indirectness imprecision N/A 0 0 HR 0. (0. to 0.) P=0.0 - HIGH OS years (sub-grouped by staging - all data). Follow-up -0 months. Winter-Roach et al (00) randomised trials limitations inconsistency indirectness imprecision N/A 0 00 HR 0. (0. to 0.) P=0.0 - HIGH OS years (sub-grouped by staging - optimal staging). Follow-up -0 months. Winter-Roach et al (00) randomised trials limitations inconsistency indirectness HR. serious imprecision N/A (0. to.) P= 0. - MODERATE OS years (sub-grouped by staging - sub-optimal staging. Follow-up -0 months. Winter-Roach et al (00) randomised trials limitations inconsistency indirectness imprecision N/A HR 0. (0. to 0.) P= HIGH OS 0 years (sub-grouped by risk - all). Follow-up -0 months. Winter-Roach et al (00) randomised trials limitations N/A indirectness N/A N/A - - totals not selected - N/A OS 0 years (sub-grouped by risk - low/medium risk). Follow-up -0 months. Winter-Roach et al (00) randomised trials limitations N/A indirectness N/A N/A - - not estimable - N/A OS 0 years (sub-grouped by risk - high risk). Follow-up -0 months. Winter-Roach et al (00) The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

70 No of studies Quality assessment Design Limitations Inconsistency Indirectness Imprecision Other No of patients Chemotherapy Observation Summary of findings Effect Relative (% CI) Absolute Quality randomised trials limitations N/A indirectness N/A N/A - - not estimable - N/A PFS years. Follow-up -0 months. Winter-Roach et al (00) randomised trials limitations inconsistency indirectness imprecision N/A HR 0. (0. to 0.) P= HIGH PFS years (data sub-grouped by staging - all). Follow-up -0 months. Winter-Roach et al (00) randomised trials limitations inconsistency indirectness imprecision N/A HR 0. (0. to 0.) P= HIGH PFS years (data sub-grouped by staging - optimal staging). Follow-up -0 months. Winter-Roach et al (00) randomised trials limitations inconsistency indirectness HR 0. serious imprecision N/A (0. to.) P=0. - MODERATE PFS years (data sub-grouped by staging - sub-optimal staging). Follow-up -0 months. Winter-Roach et al (00) randomised trials limitations inconsistency indirectness imprecision N/A 0 HR 0. (0.0 to 0.) P= HIGH PFS 0 years (sub-grouped by risk). Follow-up -0 months. Winter-Roach et al (00) randomised trial limitations N/A indirectness N/A N/A - - totals not selected - N/A PFS 0 years (sub-grouped by risk - low/medium risk). Follow-up -0 months. Winter-Roach et al (00) randomised trial limitations N/A indirectness N/A N/A - - not estimable - N/A The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 0 of

71 No of studies Quality assessment Design Limitations Inconsistency Indirectness Imprecision Other No of patients Chemotherapy Observation Summary of findings Effect Relative (% CI) Absolute Quality PFS 0 years (sub-grouped by risk - high risk). Follow-up -0 months. Winter-Roach et al (00) randomised trial limitations N/A indirectness N/A N/A - - not estimable - N/A DSS. Follow-up -0 months. Winter-Roach et al (00) randomised trial limitations N/A indirectness HR 0. serious imprecision N/A (0. to.) P=0.0 - MODERATE Death from ovarian cancer. Follow-up -0 months. Winter-Roach et al (00) randomised trials limitations inconsistency indirectness imprecision N/A / / RR 0. (0. to.) P=0. - HIGH 0 year cancer-specific survival, all patients. Follow-up 0. years. Trimbos et al (00) randomised trial limitations N/A indirectness N/A N/A % (-%) % (-%) HR 0. (0. to.) P=0. - HIGH 0 year cancer-specific survival, optimally staged patients. Follow-up 0. years. Trimbos et al (00) randomised trial limitations N/A indirectness N/A N/A % (-%) % (-%) HR. (0. to.0) P=0. - HIGH 0 year cancer-specific survival, non-optimally staged patients. Follow-up 0. years. Trimbos et al (00) The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

72 No of studies Quality assessment Design Limitations Inconsistency Indirectness Imprecision Other No of patients Chemotherapy Observation Summary of findings Effect Relative (% CI) Absolute Quality randomised trial limitations N/A indirectness N/A N/A 0% (-%) % (0-%) HR 0. (0. to 0.) P=0.0 - HIGH 0 year recurrence-free survival, all patients. Follow-up 0. years. Trimbos et al (00) randomised trial limitations N/A indirectness N/A N/A 0% (-%) % (-%) HR 0. (0. to 0.) P= HIGH 0 year recurrence-free survival, optimally staged patients. Follow-up 0. years. Trimbos et al (00) randomised trial limitations N/A indirectness N/A N/A % (-%) % (-%) HR 0. (0.-.) P=0. - HIGH 0 year recurrence-free survival, non-optimally staged patients. Follow-up 0. years. Trimbos et al (00) randomised trial limitations N/A indirectness N/A N/A % (-%) % (-%) HR 0.0 (0. to 0.) P= HIGH 0 year cancer-specific survival, patients with grade disease. Follow-up 0. years. Trimbos et al (00) randomised trial limitations N/A indirectness N/A N/A % (-%) % (-%) HR 0. (0.-.) P=0.0 - HIGH The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

73 No of studies Quality assessment Design Limitations Inconsistency Indirectness Imprecision Other cycles No of patients cycles Summary of findings Relative (% CI) Effect Absolute Quality Overall death rate years. cycles vs. cycles. Follow-up. years. Bell et al (00) randomised trial serious limitation N/A indirectness serious imprecision N/A HR.0 (0. to.) P=0. LOW Rate of recurrence years. cycles vs. cycles. Follow-up. years. Bell et al (00) randomised trial serious limitation N/A indirectness serious imprecision N/A HR 0. (0. to.) P=0. LOW Rate of recurrence. cycles vs. cycles. Follow-up months. Serous tumours. Chan et al. (00) randomised trial serious limitation N/A indirectness serious imprecision N/A 0.%.% HR 0. (0. to 0.) P=0.00 LOW Rate of recurrence. cycles vs. cycles. Follow-up months. Non-serous tumours. Chan et al. (00) 0 randomised trial Footnotes The % confidence interval spans the line of no effect and exceeds the limits of both <0. x the effect size (0.) and >. x the effect size (.). The result suggests no significant difference between comparators. The % confidence interval spans the line of no effect and exceeds the limits of both <0. x the effect size (0.0) and >. x the effect size (0.). The result suggests no significant difference between comparators. The % confidence interval spans the line of no effect and exceeds the limits of both <0. x the effect size (0.) and >. x the effect size (.0). This may due to low sample number. The result suggests no significant difference between comparators. There were few details of the randomisation allocation or assessment blinding methodology given. The % confidence interval spans the line of no effect and exceeds the limits of both <0. x the effect size (0.) and >. x the effect size (.). The result suggests no significant difference between comparators. The % confidence interval spans the line of no effect and exceeds the limits of both <0. x the effect size (0.) and >. x the effect size (0.). The result suggests no significant difference between comparators. serious limitation N/A indirectness serious imprecision N/A.%.% HR 0. (0.0 to.) P=0.0 LOW The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

74 Recommendations Do not offer adjuvant chemotherapy to women who have had optimal surgical staging and have low-risk stage I disease (grade or, stage Ia or b). Discuss the possible benefits and side effects of adjuvant chemotherapy with women who have had suboptimal surgical staging and have stage I disease. Offer women with high-risk stage I disease (grade or stage Ic) six cycles of adjuvant carboplatin (but see also next recommendation). Consider three cycles of adjuvant carboplatin plus paclitaxel for women with high-risk stage I disease (grade or stage Ic) if they are prepared to accept treatment of shorter duration but increased toxicity. Linking evidence to recommendations Interventions that improve the likelihood of disease free survival are very important, but that benefit needs to be weighed against the morbidity and effects on overall quality of life. The GDG noted that there was some evidence suggesting adjuvant chemotherapy in stage I disease could reduce the risk of relapse and death from ovarian cancer. This evidence was limited and of varying quality on the basis of study quality assessed according to GRADE. The GDG was aware that there was a lack of data on both the toxicity associated with adjuvant chemotherapy and on how this affected quality of life. In women whose risk of relapse was small the GDG felt the adverse effects and costs of adjuvant treatment would significantly outweigh any benefit from treatment and therefore did not recommend adjuvant chemotherapy. The GDG was also aware that different women might place different personal value on the short-term adverse effects of treatment as well as on the possible long-term benefits. Therefore discussion of treatment options, as well as the option of no treatment was important. The GDG noted that single agent platinum-based therapy, using cycles of carboplatin, had demonstrated a survival benefit in women with early stage ovarian cancer. They were also aware that combination therapy had been shown to be more toxic than monotherapy. In addition, there was no evidence that combination therapy was any more effective than monotherapy in early stage disease. The GDG therefore decided to recommend cycles of adjuvant carboplatin for most women, in keeping with current standard practice. However the GDG acknowledged that combination therapy could be useful for those women who were prepared to accept a shorter treatment duration, but with increased toxicity. They were aware of evidence that cycles of combination therapy was less toxic than cycles, therefore they decided to recommend that cycles of paclitaxel plus carboplatin be considered as an option for women with ovarian cancer. This clinical question was considered a low priority for health economic evaluation because of the small patient numbers involved. References Optimal surgical staging constitutes midline laparotomy to allow thorough assessment of the abdomen and pelvis; a total abdominal hysterectomy, bilateral salpingo-oophorectomy and infracolic omentectomy; biopsies of any peritoneal deposits; random biopsies of the pelvic and abdominal peritoneum and retroperitoneal lymph node assessment [Winter Roach et al. (00) Adjuvant (post-surgery) chemotherapy for early stage epithelial ovarian cancer. Cochrane Database of Systematic Reviews 00, Issue : CD000} In UK clinical practice, paclitaxel is usually provided in combination with carboplatin (rather than with cisplatin) for treating ovarian cancer, because of the well established lower toxicity of this combination. However, paclitaxel in combination with carboplatin does not have a UK marketing authorisation for treating ovarian cancer, so informed consent should be obtained and documented. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

75 0 0 0 Bell, J., M. F. Brady, R. C. Young, J. Lage, J. L. Walker, K. Y. Look, G. S. Rose, N. M. Spirtos, and Gynaecologic Oncology Group (00). Randomized phase III trial of three versus six cycles of adjuvant carboplatin and paclitaxel in early stage epithelial ovarian carcinoma: a Gynaecologic Oncology Group study. Gynaecol Oncol 0: -. Chan JK., Munro EG., Cheung MK., Husain A., Teng NN., Berek JS and Osann K (00) Association of lymphadenectomy and survival in stage I ovarian cancer patients. Obstetrics and Gynecology 0: -. Chan JK, Tian C, Fleming GF, Monk BJ, Herzog TJ, Kapp DS et al.. (00). The potential benefit of vs. cycles of chemotherapy in subsets of women with early-stage high-risk epithelial ovarian cancer: an exploratory analysis of a Gynecologic Oncology Group study. Gynecol Oncol (): 0-0. Kim HS, Ju W, Jee BC, Kim YB, Park NH, Song YS et al.. (00) Systematic lymphadenectomy for survival in epithelial ovarian cancer a meta-analysis. Int J Gynecol Cancer 0():0-. Maggioni A., Benedetti PP., Dell'Anna T., Landoni F., Lissoni A., Pellegrino A., Rossi RS., Chiari S., Campagnutta E., Greggi S., Angioli R., Manci N., Calcagno M., Scambia G., Fossati R., Floriani I., Torri V., Grassi R and Mangioni C (00) Randomised study of systematic lymphadenectomy in patients with epithelial ovarian cancer macroscopically confined to the pelvis. Br J Cancer : - 0 National Institute for Health and Clinical Excellence (00) Guidance on the use of paclitaxel in the treatment of ovarian cancer. NICE technology appraisal guidance. London: National Institute for Health and Clinical Excellence. Suzuki S, Kajiyama H, Shibata K, Ino K, Nawa A, Sakakibara K, Matsuzawa K, Takeda A, Kinoshita Y, Kawai M, Nagasaka T and Kikkawa F (00). Is there any association between retroperitoneal lymphadenectomy and survival benefit in ovarian clear cell carcinoma patients? Ann.Oncol. : -. Trimbos B, Timmers P, Pecorelli S, Coens C, Ven K, van der Burg M et al.. (00). Surgical staging and treatment of early ovarian cancer: long-term analysis from a randomized trial. J Natl Cancer Inst 0(): -. Winter-Roach BA., Kitchener HC and Dickinson HO (00). Adjuvant (post-surgery) chemotherapy for early stage epithelial ovarian cancer. Cochrane DB Sys Rev 00, Issue. Art. No: CD000. DOI: 0.00/.CD000.pub. Yang X., Hou M., Yang K., Wang H., Peng., Cao Z and Mingrong X. (00) Prognosis in epithelial ovarian cancer: clinical analysis of pelvic and para-aortic lymphadenectomy. Chinese-German Journal of Clinical Oncology (): -. Yokoyama Y., Sakamoto T., Sato S and Saito Y (). Evaluation of cytoreductive surgery with pelvic and para-aortic lymphadenectomy and intermittent cisplatin-based combination chemotherapy for improvement of long-term survival in ovarian cancer. Eur J Gynaecol Oncol 0: -. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

76 Management of advanced (stage II-IV) ovarian cancer The two objectives of this chapter were:. to assess the role of surgery in the treatment of women with advanced stage (II-IV) ovarian cancer and to determine the optimal timing of surgery within the treatment pathway. to determine the clinical benefits and toxicity of intraperitoneal chemotherapy given as part of the first-line management of advanced stage (II-IV) ovarian cancer.. The value of primary surgery Surgery can be either primary (performed for the first time, either before, during or after chemotherapy) or secondary (performed after primary surgery). Secondary surgery can be sub-classified into either being early, when performed during chemotherapy (usually termed interval debulking surgery or IDS) or late when performed after primary chemotherapy (also called second-look laparotomy). Historically, surgery has been an integral part of treating ovarian cancer, and before the advent of radiotherapy and chemotherapy, the only treatment. This historical fact accounts for why surgery came to occupy the position it does without formal scrutiny. Surgery alone can be curative when cancer is confined to the ovaries but this is not true for the majority of women with ovarian cancer; for whom surgery can only be cytoreductive (debulking). The value of surgery in these circumstances is not fully understood. It was only with the introduction of active chemotherapy (in particular cisplatin) that aggressive cytoreductive surgery was undertaken, even when it was clear at the outset that all the disease could not be removed. The beneficial effects of cytoreductive surgery are only seen in conjunction with active chemotherapy and the independent contribution of surgery in this context remains to be established. There are many studies that have shown a negative association between the amount of residual disease after surgery and outcome, but these studies are retrospective and uncontrolled (Griffiths ; Parker et al., 0; Hacker et al., ; Wharton et al., ; Lyngstadaas 00). Therefore, although the amount of disease remaining at the end of the operation is a powerful adverse prognostic factor, it cannot be assumed that this association is one of cause and effect. It is possible, for example, that cancers that are more difficult to resect have a different tumour biology and responsiveness to chemotherapy. Similarly, a number of studies (Junor et al., ) have shown an association between the type of surgeon and outcome; thus women presenting (electively) to gynaecological oncologists fare better than those operated on by general surgeons (women often presenting as emergencies with intestinal obstruction). The observed survival advantage of being operated on by a gynaecological oncologist may be because better rates of optimal cytoreduction were achieved but it could also be that the patient groups were very different. Only adequately designed and conducted prospective RCTs would effectively address these confounding variables. Clinical question: What is the effectiveness of surgery in the primary management of women with ovarian cancer who will receive chemotherapy? Clinical evidence The evidence for this topic was limited and consisted of two Cochrane systematic reviews and two small randomised controlled trials (RCTs) which dealt with different aspects of The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

77 0 0 0 surgery (Table.). The total number of women across studies was,0 and all but stage I disease was represented. None of the studies addressed patient quality of life. Morrison et al., (00) conducted a Cochrane review of chemotherapy versus surgery for the initial treatment of advanced ovarian cancer. Despite an extensive search of the literature, the authors identified only one small RCT which had randomised women to receive either one cycle of chemotherapy followed by embolisation of the ovarian artery, debulking surgery and adjuvant chemotherapy or debulking surgery and adjuvant chemotherapy only. There was no statistically significant difference in median overall survival ( months [%CI:.-. months] versus months [%CI:.-. month]) (P>0.0)) between treatments. The chemo-embolisation arm did experience less surgery related morbidity but no other adverse events were reported. Tangitjamol et al., (00) reviewed three RCTs in which women with ovarian cancer who had undergone sub-optimal primary surgery were randomised to chemotherapy with interval debulking surgery (IDS) or chemotherapy without IDS. There was significant between studies heterogeneity and so the authors performed sub-group analyses. They concluded that if women had received their primary surgery from a general surgeon, as opposed to a gynaecological oncologist, or had received less extensive surgery, then IDS showed a marginal survival benefit (RR=0. [%CI: 0.-0.] P=0.00). There was no statistically significant difference between study arms in terms of either adverse events or quality of life. Nicoletto et al., () randomised 0 women with ovarian cancer, who had an apparently complete clinical response to primary surgery and adjuvant chemotherapy, to either secondlook surgery or a watch and wait policy. After a mean follow-up of 0 months the authors could demonstrate no significant difference in overall survival (HR=0. [%CI: 0.-.] P=0.) even though patients with a positive second-look surgery were subsequently treated with non cross-reactive chemotherapy. Luesley et al., () recruited women with ovarian cancer who had received primary surgery (but were left with residual disease) and adjuvant cisplatin, randomising them to receive either second-look surgery followed by chemotherapy with chlorambucil or pelvic irradiation. A third group received chemotherapy only. With an average follow-up of months, there was no significant difference in median overall survival between the two surgical groups ( months [%CI: - months] versus months [%CI: - months)] P=0.)) or between the surgery plus chemotherapy group versus the chemotherapy only group ( months [%CI: - months] versus months [%CI: - months] P=0.)). The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

78 Table. GRADE profile: What is the effectiveness of surgery in the primary management of women with ovarian cancer who will receive chemotherapy? No of studies Design Limitations Quality assessment Inconsiste ncy Indirectness Imprecision Other Chemotherapy before surgery Time in months Chemotherapy after surgery Summary of findings Relative (% CI) Effect Absolute Quality Mean OS (P>0.0). Follow-up months (range: - months) Liu et al., 00 (in Morrison et al. 00) RCT serious limitations N/A indirectness serious imprecision N/A. (%CI:.-.). (%CI:.-.) - - LOW Median OS (P>0.0). Follow-up months (range: - months) Liu et al., 00 (in Morrison et al. 00) RCT serious limitations N/A indirectness serious imprecision N/A (%CI:.-.) (%CI:.-.) - - LOW Median DFI (P>0.0). Follow-up months (range: - months) Liu et al., 00 (in Morrison et al. 00) RCT serious limitations N/A indirectness serious imprecision N/A. (no %CI). (no %CI) - - LOW Overall survival (χ =.; P>0.0). Follow-up months (range: - months) Liu et al., 00 (in Morrison et al. 00) 0 RCT serious limitations N/A indirectness serious imprecision N/A LOW Footnotes This was a non-english language study that had not apparently been translated by the Cochrane reviewers. Although the original study authors stated that they had randomised patients, there were no details of randomisation or allocation and blinding of outcome assessors was not mentioned. Intention to treat (ITT) analysis was used but treatment withdrawals were no discussed. The Kaplan Meier plot and tables accompanying the text of Liu et al., (00) were not accessible and may have included more data with regard to survival. However this was a low patient number trial. Patients: women with stage III (actually II) or IV EOC; Intervention: neoadjuvant intra-arterial chemo ( cycle), ovarian artery embolisation then primary surgery followed by adjuvant i.v. chemo ( cycles) (n=); Control: primary surgery followed by adjuvant i.v. chemo ( cycles) (n=). The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

79 No of studies Quality assessment Design Limitations Inconsistency Indirectness Imprecision Other Interval debulking surgery Patients No interval debulking surgery Summary of findings Relative (% CI) Effect Absolute Quality Risk of death (P=0.0) (if surgery was performed by general surgeons). Follow-up - months. Tangjitgamol et al., 00 RCT limitations inconsistency indirectness imprecision N/A 0 RR=0. (0.-0.) - HIGH Risk of death (P=0.) (if surgery was less extensive or performed by gynaecological surgeons). Follow-up - months. Tangjitgamol et al., 00 RCT limitations N/A indirectness imprecision N/A 0 RR=0. (0.-.) - HIGH Toxic reactions to chemotherapy (P=0.). Follow-up - months. Tangjitgamol et al., 00 RCT limitations inconsistency indirectness serious imprecision N/A / /0 RR=. (0.-.) fewer per 00 Footnotes The three included studies in this systematic review were described by the authors as having given sufficient details of randomisation and allocation but blinding of treatment assessors was not described. All studies used intention to treat (ITT) analysis. The original pooled data for survival from the three included studies showed significant heterogeneity (I =%) and the authors addressed this by stratifying data by surgical speciality, as shown in the table. The confidence interval around the estimate of effect spans (the line of no effect) and the limits for appreciable harm and appreciable benefit. MODERATE The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

80 No of studies Quality assessment Design Limitations Inconsistency Indirectness Imprecision Other nd look surgery Patients Watchful waiting Summary of findings Relative (% CI) Effect Absolute Quality Overall survival (χ =0.; P=0.). Follow-up ~0 months. Nicoletto et al., RCT serious limitations N/A No of studies indirectness serious imprecision N/A HR=0. (0.-.) Footnotes This study did not demonstrate adequate details of randomisation, allocation or blinding of treatment assessors. The study used intention to treat (ITT) analyses. The confidence interval is wide and crosses the line of no effect as well as exceeding limits for appreciable harm and appreciable benefit. This is probably due to the low patient number Quality assessment Design Limitations Inconsistency Indirectness Imprecision Other [A] nd look surgery then chemotherapy Patients Summary of findings [B] nd look surgery then radiotherapy [C] Chemotherapy - LOW Quality Median survival (A vs. B: χ =0.; P=0.; A vs. C: χ =0.; P=0.). Follow-up months (range: - months). Luesley et al., 0 0 RCT very serious limitations N/A indirectness months very serious imprecision N/A (%CI: - months) N=/ months (%CI: - months) N=/ months (%CI: - months) N=/ Footnotes This study did not demonstrate adequate details of randomisation, allocation, blinding of treatment assessors or intention to treat (ITT) analysis. The comparison of Group A vs. Group C may be unsafe since, on the Kaplan Meier plot shown, the lines representing each population cross several times. The statistics (chi square and P value) from Groups A vs. B and A vs. C are identical which may be accurate or not. The study is probably underpowered to detect a significant difference between study arms. VERY LOW The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 0 of

81 Research recommendation Research should be undertaken to determine the effectiveness of primary surgery for women with advanced ovarian cancer whose tumour cannot be fully excised. Linking evidence to recommendations The GDG placed a high value on the outcomes of survival and morbidity. They noted that the evidence, using the GRADE quality assessment tool, concerning surgery was limited, of poor quality, contradictory and open to interpretation. Therefore the GDG made recommendations for further research into the effectiveness of surgery. This clinical question was considered a low priority for economic analysis because although the number of patients involved could potentially be large, there was considerable uncertainty over the health benefits of performing surgery, due to a lack of RCT data.. Intraperitoneal chemotherapy Ovarian cancer commonly involves the peritoneal surfaces of the intra-abdominal cavity without distant metastatic spread. Efforts to directly target small volume tumour deposits have included the use of intra-peritoneal stripping, monoclonal antibodies, radionuclides and intraperitoneal chemotherapy. The most promising of these strategies is intraperitoneal chemotherapy and several studies have shown a moderate improvement in disease free and overall survival. Most of these trials are dated, being carried out in the early 0s, involving agents such as cisplatin and cyclophosphamide, now considered inferior to the current agents of carboplatin with or without paclitaxel. Two more recent trials have reignited the interest in intraperitoneal chemotherapy and confirmed the feasibility of administering paclitaxel by this route. Both trials reported significant immediate toxicities and further research is urgently needed. Clinical question: For women with ovarian cancer, is intraperitoneal chemotherapy effective in primary management? Clinical evidence The evidence for this topic comprises two high quality systematic reviews (Jaaback and Johnson, 00; Elit et al., 00) and one randomised controlled trial (RCT) (Wenzel et al., 00) (Table.). Between them, these studies reported on all the outcomes of interest. The two systematic reviews included meta-analyses of data from the same RCTs but both reviews were appraised because the authors reported different survival outcomes. The majority of trial data derived from the United States of America and all the studies compared the use of standard intravenous chemotherapy with chemotherapy regimens incorporating a component of intra-peritoneal drug delivery for the first line adjuvant treatment of primary ovarian cancer. High quality evidence from pooled data from up to eight trials suggested that chemotherapy given directly into the peritoneal cavity as part of adjuvant treatment, may significantly reduce the risk of death (HR: 0.0 [%CI: ] P=0.000) and disease recurrence (HR: 0. [%CI: ] P=0.000) an effect also seen after five years of follow-up (RR of death: 0. [%CI: 0.-0.] P=0.00; RR of disease progression: 0. [%CI: 0.] P=0.0). However, incidences of pain, fever, fatigue, hearing loss, infection and gastrointestinal and metabolic effects occurred up to eight times more frequently in women receiving intra-peritoneal chemotherapy. The one exception to this observation was the incidence of cardiovascular effects which were not significantly different between study arms. The evidence about haematological, pulmonary, renal and neurological adverse The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

82 effects was too poor in quality to allow conclusions to be drawn about the relative contribution of the drug delivery route. Health-related quality of life was measured in one trial and found to be significantly worse for women receiving intra-peritoneal chemotherapy in the early days of treatment and shortly ( to weeks) after all study treatment, but a difference between study arms was not apparent after one year of follow-up. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

83 Table. GRADE profile: For women with ovarian cancer, is intra-peritoneal chemotherapy effective in primary management Quality assessment No of patients Summary of findings Effect No of studies Design Limitations Inconsistency Indirectness Imprecision Other IP chemotherapy IV chemotherapy Relative (% CI) Absolute Quality Time to death (follow-up to months ). Effect size < favours intraperitoneal chemotherapy. Jaaback and Johnson (00). randomised trials serious, inconsistency indirectness imprecision N/A HR 0.0 (0. to 0.) P=0.000 MODERATE Time to death (high quality studies only) (follow-up to months ). Effect size < favours intraperitoneal chemotherapy. Jaaback and Johnson (00). randomised trials limitations, inconsistency indirectness imprecision N/A 0 HR 0. (0. to 0.) P=0.000 HIGH Time to recurrence (follow-up to months ). Effect size < favours intraperitoneal chemotherapy. Jaaback and Johnson (00). randomised trials limitations, inconsistency indirectness imprecision N/A HR 0. (0. to 0.) P=0.000 HIGH Time to recurrence (high quality studies only) (follow-up to months ). Effect size < favours intraperitoneal chemotherapy. Jaaback and Johnson (00). randomised trials limitations, inconsistency indirectness imprecision N/A HR 0. (0. to 0.) P=0.000 HIGH Survival (risk of death) years (follow-up to months ). Effect size < favours intraperitoneal chemotherapy. Elit et al. (00). randomised trials limitations inconsistency indirectness imprecision N/A / (.%) / (.%) RR 0. (0. to 0.) P=0.00 fewer per 00 (from 0 fewer to fewer) HIGH Progression-free survival (risk of progression) at years (follow-up to months ). Effect size < favours intraperitoneal chemotherapy. Elit et al. (00). randomised trials limitations inconsistency indirectness imprecision N/A / (%) / (.%) RR 0. (0. to 0.) P=0.0 fewer per 00 (from fewer to fewer) HIGH The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

84 Quality assessment No of patients Summary of findings Effect No of studies Design Limitations Inconsistency Indirectness Imprecision Other IP chemotherapy IV chemotherapy Relative (% CI) Absolute Quality Adverse effects anaemia. Effect size < favours intraperitoneal chemotherapy. Jaaback and Johnson (00). randomised trials serious inconsistency indirectness serious 0 N/A / (0.%) / (.%) RR 0. ( to.) P=0.0 fewer per 00 (from more to more) LOW Adverse effects thrombocytopenia. Effect size < favours intraperitoneal chemotherapy. Jaaback and Johnson (00). randomised trials serious very serious, indirectness serious 0 N/A / (.%) / (.%) RR. (0. to.0) P=0. more per 00 (from fewer to more) VERY LOW Adverse effects leukopenia. Effect size < favours intraperitoneal chemotherapy. Jaaback and Johnson (00). randomised trials serious very serious, indirectness imprecision N/A / (%) / (.%) RR 0. (0. to.) P=0. fewer per 00 (from fewer to 0 more) VERY LOW Adverse effects renal. Effect size < favours intraperitoneal chemotherapy. Jaaback and Johnson (00). randomised trials serious serious, indirectness imprecision N/A / (.%) / (.%) RR. (0. to.) P=0. more per 00 (from 0 fewer to more) LOW Adverse effects pulmonary. Effect size < favours intraperitoneal chemotherapy. Jaaback and Johnson (00). randomised trials limitations serious, indirectness imprecision N/A 0/ (.%) / (.%) RR. (0. to.) P=0. more per 00 (from fewer to 0 more) MODERATE Adverse effects cardiovascular. Effect size < favours intraperitoneal chemotherapy. Jaaback and Johnson (00). randomised trials limitations inconsistency indirectness imprecision N/A /0 (.%) / (.%) RR. (0. to.0) P=0.0 more per 00 (from 0 fewer to more) HIGH Adverse effects fever. Effect size < favours intraperitoneal chemotherapy. Jaaback and Johnson (00). The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

85 Quality assessment No of patients Summary of findings Effect No of studies Design Limitations Inconsistency Indirectness Imprecision Other IP chemotherapy IV chemotherapy Relative (% CI) Absolute Quality randomised trials limitations 0 inconsistency indirectness imprecision N/A / (.%) / (.%) RR. (. to.0) P=0.00 more per 00 (from more to more) HIGH Adverse effects fatigue. Effect size < favours intraperitoneal chemotherapy. Jaaback and Johnson (00). randomised trials limitations inconsistency indirectness imprecision N/A /0 (.%) / (.%) RR. (. to.) P=0.000 more per 00 (from more to more) HIGH Adverse effects gastrointestinal. Effect size < favours intraperitoneal chemotherapy. Jaaback and Johnson (00). randomised trials serious serious indirectness imprecision N/A 0/ (%) / (.%) RR.0 (. to.) P=0.00 more per 00 (from more to more) LOW Adverse effects infection. Effect size < favours intraperitoneal chemotherapy. Jaaback and Johnson (00). randomised trials limitations inconsistency indirectness imprecision N/A /0 (0%) / (.%) RR. (. to.) P=0.000 more per 00 (from more to more) HIGH Adverse effects metabolic. Effect size < favours intraperitoneal chemotherapy. Jaaback and Johnson (00). randomised trials limitations inconsistency indirectness imprecision N/A /0 (.%) / (.%) RR. (. to.) P< more per 00 (from more to more) HIGH Adverse effects neurological. Effect size < favours intraperitoneal chemotherapy. Jaaback and Johnson (00). randomised trials limitations serious, indirectness serious 0 N/A 0/ (.%) /0 (.%) RR. (0. to.0) P=0. more per 00 (from fewer to more) LOW Adverse effects pain. Effect size < favours intraperitoneal chemotherapy. Jaaback and Johnson (00). randomised trials limitations inconsistency indirectness imprecision N/A / (.%) / (.%) RR. (. to.) P< more per 00 (from more to more) HIGH The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

86 Quality assessment No of patients Summary of findings Effect No of studies Design Limitations Inconsistency Indirectness Imprecision Other IP chemotherapy IV chemotherapy Relative (% CI) Absolute Quality Adverse effects hearing loss. Effect size < favours intraperitoneal chemotherapy. Jaaback and Johnson (00). randomised trials limitations inconsistency indirectness imprecision N/A / (.%) / (.%) RR 0. (0. to 0.) P=0.0 fewer per 00 (from 0 fewer to fewer) HIGH QOL at baseline (FACT-G) (FACT-O measured from 0 to units. Higher values indicate better QOL). MD compares IV to IP chemotherapy. Wenzel et al. (00). randomised trial limitations inconsistency indirectness imprecision N/A 0 - MD. higher (0. to. higher) P=0.0 HIGH QOL at baseline (FACT-O subscale) (FACT-O measured from 0 to units. Higher values indicate better QOL). MD compares IV to IP chemotherapy. Wenzel et al. (00). randomised trial limitations inconsistency indirectness imprecision N/A 0 - MD. higher (0. to. higher) P=0.00 HIGH QOL before cycle (FACT-G) (FACT-O measured from 0 to units. Higher values indicate better QOL). MD compares IV to IP chemotherapy. Wenzel et al. (00). randomised trial limitations inconsistency indirectness imprecision N/A - MD. higher (. to. higher) P<0.00 HIGH QOL before cycle (FACT-O subscale) (FACT-O measured from 0 to units. Higher values indicate better QOL). MD compares IV to IP chemotherapy. Wenzel et al. (00). randomised trial limitations inconsistency indirectness imprecision N/A - MD. higher (. to. higher) P<0.00 HIGH QOL - weeks after treatment (FACT-G) (FACT-O measured from 0 to units. Higher values indicate better QOL). MD compares IV to IP chemotherapy. Wenzel et al. (00). randomised trial limitations inconsistency indirectness imprecision N/A - MD. higher (. to. higher) P=0.00 HIGH The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

87 No of studies Quality assessment Design Limitations Inconsistency Indirectness Imprecision Other No of patients IP chemotherapy IV chemotherapy Summary of findings Effect Relative (% CI) Absolute Quality QOL - weeks after treatment (FACT-O subscale) (FACT-O measured from 0 to units. Higher values indicate better QOL). MD compares IV to IP chemotherapy. Wenzel et al. (00). randomised trial limitations inconsistency indirectness imprecision N/A - MD. higher (0. to. higher) P=0.0 HIGH QOL year after treatment (FACT-G) (measured from 0 to units. Higher values indicate better QOL). MD compares IV to IP chemotherapy. Wenzel et al. (00). randomised trial limitations inconsistency indirectness imprecision N/A 0 - MD 0. higher (. lower to. higher) P=0. HIGH QOL year after treatment (FACT-O subscale) (measured from 0 to units. Higher values indicate better QOL). MD compares IV to IP chemotherapy. Wenzel et al. (00). 0 0 randomised trial limitations inconsistency indirectness imprecision N/A 0 - MD 0. higher (. lower to. higher) P=0. Footnotes / trials reported duration of follow-up which in trials was stated to be >0 months. The review authors reported and assessed the allocation method, concealment, assessor blinding and intention-to-treat for all studies. On this basis they judged studies to be 'good', studies as fair and studies as poor in quality. Details of loss to follow-up are not reported for individual studies or overall. For this outcome, papers have been graded 'good', as 'fair' and as 'poor'. For this outcome, papers have been graded 'good' and as 'fair'. For this outcome, papers have been graded 'good', as 'fair' and as 'poor'. For this outcome, papers have been graded 'good' and as 'fair'. For this outcome, paper has been graded 'good', as 'fair' and as 'poor'. For this outcome, papers have been graded 'good', as 'fair' and as 'poor'. For this outcome, papers have been graded 'good'. 0 For this outcome, papers have been graded 'good' and as 'fair'. For this outcome, papers have been graded 'good', as 'fair' and as 'poor'. For this outcome, papers have been graded 'good' and as poor. High levels of between studies heterogeneity in adverse effects outcomes are explained adequately in the review discussion highlighting the fact that different drugs, doses and regimes were used across studies. Also, / of the studies used extremely high doses of chemotherapy in the intraperitoneal chemotherapy which increased the likelihood of adverse events. The authors conclude that for leukopenia, thrombocytopenia, renal, neurological and pulmonary outcomes, data pooling (although undertaken) could be considered inappropriate. Between studies heterogeneity was measured at 0%. Between studies heterogeneity was measured at 0%. Between studies heterogeneity was measured at %. Between studies heterogeneity was measured at %. Between studies heterogeneity was measured at %. HIGH The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

88 The % confidence interval crosses the line of no effect plus the lower value of the interval is <0. and/or upper value >.. But the event rate is <% so study quality is not downgraded. 0 The % confidence interval crosses the line of no effect plus the lower value of the interval is <0. and/or upper value >.. But the event rate is >% so study quality is downgraded Calculated as a raw difference for data before randomisation and adjusted mean difference for all time points thereafter. NB. FACT-O score = scores of FACT-O subscale & FACT-G combined. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

89 Recommendation Do not offer intraperitoneal chemotherapy to women with ovarian cancer (any stage) except as part of a clinical trial. Linking evidence to recommendations The GDG placed a high value on improving the outcomes of disease-free and overall survival, both of which were shown to benefit from the use of intra-peritoneal chemotherapy compared to standard intravenous chemotherapy. However, the GDG recognised that intra-peritoneal chemotherapy was associated with more toxicity/adverse events than standard intravenous chemotherapy and that one study had shown health-related quality of life to be adversely affected by intraperitoneal chemotherapy in the short term. The GDG also recognised that the administration of intra-peritoneal chemotherapy was more complex and more expensive than that for standard intravenous chemotherapy. Although there was high-quality evidence (assessed according to GRADE analysis) on the use of intra-peritoneal chemotherapy, the GDG noted that the studies investigated historical drug regimens and did not investigate intra-peritoneal administration of drugs given intra-venously in current standard UK regimens. There was also a lot of heterogeneity in the studies making it difficult to draw robust conclusions from the evidence. In addition, only one study presented quality of life data and so it was difficult to know if these data were representative. Based on this the GDG did not feel able to recommend the use of intra-peritoneal chemotherapy outside of clinical trials. This clinical question was not considered to be a high priority for health economic evaluation due to a relatively small patient group and a lack of evidence related to current chemotherapy agents.. Chemotherapy regimens The following recommendations are taken from Guidance on the use of paclitaxel in the treatment of ovarian cancer, NICE technology appraisal guidance (NICE, 00). These recommendations refer to both early and advanced disease and should be read in conjunction with chapter. Recommendations It is recommended that paclitaxel in combination with a platinum-based compound or platinum-based therapy alone (cisplatin or carboplatin) are offered as alternatives for first-line chemotherapy (usually following surgery) in the treatment of ovarian cancer. The choice of treatment for first-line chemotherapy for ovarian cancer should be made after discussion between the responsible clinician and the patient about the risks and benefits of the options available. In choosing between treatment with a platinum-based compound alone or paclitaxel in combination with a platinum-based compound, this discussion should cover the side-effect profiles The recommendations from NICE technology appraisal guidance will be incorporated into this guideline subject to a technology appraisal review proposal agreement. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

90 of the alternative therapies, the stage of the woman s disease, the extent of surgical treatment of the tumour, and disease-related performance status. When relapse occurs after an initial (or subsequent) course of first-line chemotherapy, additional courses of treatment with the chosen chemotherapy regimen (re-challenge therapy) should be considered if the initial (or previous) response has been adequate in extent and duration. Once the tumour fails to respond adequately to the chosen first-line regimen, different treatment options should be considered as part of second-line therapy (see next recommendation). Paclitaxel is not recommended as second-line (or subsequent) therapy in women with ovarian cancer who have received the drug as part of their first-line treatment. For women who have not received paclitaxel as part of first-line treatment, it should be considered as one option alongside other drugs licensed for second-line treatment of ovarian cancer. Only oncologists specialising in ovarian cancer should supervise the provision of chemotherapy in ovarian cancer. Linking evidence to recommendations These recommendations are from Guidance on the use of paclitaxel in the treatment of ovarian cancer, NICE technology appraisal guidance (NICE 00). They were formulated by the technology appraisal and not by the guideline developers. They have been incorporated into this guideline in line with NICE procedures for developing clinical guidelines, and the evidence to support these recommendations can be found at References Elit L., Oliver TK., Covens A., Kwon J., Fung MF., Hirte HW and Oza AM. (00) Intraperitoneal chemotherapy in the first-line treatment of women with stage III epithelial ovarian cancer: a systematic review with metaanalyses. Cancer 0(): -0. Griffiths CT () Surgical resection of tumour bulk in the primary treatment of ovarian carcinoma. Natl Cancer Inst Mono : 0 0. Hacker NF, Berek JS, Lagasse LJ, Neilburg RK, Elasoff RM () Primary cytoreductive surgery for epithelial ovarian cancer. Obstetrics and Gynecology Apr; (): -0. Jaaback K and Johnson N (00). Intraperitoneal chemotherapy for the initial management of primary epithelial ovarian cancer. Cochrane Database of Systematic Reviews 00 Issue. Art. No. CD000. Junor EJ, Hole DJ, Gillis CR () Management of ovarian cancer: referral to a multidisciplinary team matters. Br J Cancer 0: 0. Liu EL and Mi RR. (00) Neoadjuvant intraarterial chemotherapy and embolization in treatment of advanced ovarian epithelial carcinoma. Chinese Medical Journal (Engl) (0):. Luesley D, Blackledge G, Kelly K, Wade-Evans T, Fielding J, Lawton F, Hilton C, Rollason T, Jordan J, Latief T, Chan KK. () Failure of second-look laparotomy to influence survival in epithelial ovarian cancer. Lancet (): -0 Lyngstadaas A, Ekanger R, Hagen B, et al. (00) Primary treatment of ovarian cancer. Tidsskr Nor Laegeforen :. Morrison J, Swanton A, Collins S, Kehoe S. (00) Chemotherapy versus surgery for initial treatment in advanced ovarian epithelial cancer. Cochrane Database of Systematic Reviews 00, Issue. Art. No.: CD00. DOI: 0.00/.CD00.pub. National Institute for Health and Clinical Excellence (00) Guidance on the use of paclitaxel in the treatment of ovarian cancer. NICE technology appraisal guidance. London: National Institute for Health and Clinical Excellence. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 0 of

91 0 Nicoletto MO, Tumolo S, Talamini R, Salvagno L, Franceschi S, Visona E, Marin G, Angelini F, Brigato G, Scarabelli C, Carbone A, Cecchetto A, Prosperi A, Rosabian A, Giusto M, Cima GP, Morassut S, Nascimben O, Vinante O, Fiorentino MV. () Surgical second look in ovarian cancer: a randomized study in patients with laparoscopic complete remission--a Northeastern Oncology Cooperative Group-Ovarian Cancer Cooperative Group Study. J Clin Oncol. (): -. Parker RT, Parker CH, Willbanks GD (0) Cancer of the ovary: survival studies based upon operative therapy, chemotherapy and radiotherapy. Am J Obstet Gynecol 0;-. Tangjitgamol S, Manusirivithaya S, Laopaiboon M, Lumbiganon P. (00) Interval debulking surgery for advanced epithelial ovarian cancer. Cochrane Database of Systematic Reviews 00, Issue. Art. No.: CD000. DOI: 0.00/.CD000.pub. Wenzel LB., Huang HQ., Armstrong DK., Walker JL and Cella D. (00) Health-related quality of life during and after intraperitoneal versus intravenous chemotherapy for optimally debulked ovarian cancer: A Gynecologic Oncology Group Study. J Clin Oncol (): -. Wharton JT, Edwards CL, Rutledge FN () Long term survival after chemotherapy for advanced ovarian epithelial carcinoma. Am J Obstet Gynecol ;-00. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

92 Support needs for women with newly diagnosed ovarian cancer Previous guidance on Improving outcomes in gynaecological cancers (Department of Health, ) made recommendations on a number of patient perspectives related to gynaecological cancer. These included the need for effective communication, delivery of relevant and timely information, and psychosocial and psychosexual support and counselling. In addition NICE service guidance Improving palliative and supportive care for adults with cancer (NICE, 00) has set standards to ensure that patients with cancer, along with their families and carers, receive the support and care to help them cope with cancer and its treatment at all stages. Healthcare professionals involved in the care of women with ovarian cancer are expected to implement the recommendations made in Improving outcomes in gynaecological cancers (Department of Health, ) and Improving supportive and palliative care for adults with cancer (NICE, 00). Implementation of these recommendations is monitored by the National Cancer Peer Review Programme in England. This programme involves self assessment by MDTs and external reviews of teams conducted by professional peers against nationally agreed peer review measures. In Wales there is a similar process of self assessment against national minimum standards for gynaecological cancers. This section of the guideline specifically focuses on the support needs of women newly diagnosed with ovarian cancer, and the psychosocial and psychosexual issues that are particular to them. Women diagnosed with ovarian cancer have a range of information and support needs, whose types and timing are as varied as the people reporting them. These needs tend to be connected with treatment, its side effects, the disease and its prognosis, as well as issues regarding sexuality. The Department of Health guidance Improving outcomes in gynaecological cancers (Department of Health, ) included recommendations about psychosocial support and psychosexual counselling and stated that psychosocial support should be available at every stage to help patients and their families to cope with the effects of the disease and its treatment. In addition, specialist interventions should be available for women and their partners to help them to understand and cope with the effects of treatment on sexual relationships. The guidance recommends that each patient should have access to a named oncology clinical nurse specialist with counselling expertise. Clinical question: For women newly diagnosed with ovarian cancer, what support should be offered? Clinical evidence Evidence from qualitative studies suggests that most women with ovarian cancer need emotional support. Improving outcomes in gynaecological cancers The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

93 (Department of Health, ), made a series of recommendations to improve supportive care in this group. However, there is evidence from the Pathfinder study (Target Ovarian Cancer, 00) that emotional support needs still go unmet in a minority of patients. Clinical nurse specialists play an important role in emotional support for women with ovarian cancer (Jefferies, 00; Target Ovarian Cancer, 00), but there is evidence that there is variation in the workloads of nurse specialists and the resources available to them (Target Ovarian Cancer, 00). In the Pathfinder study only % of the women who responded were given contact details for a clinical nurse specialist at the time of diagnosis. Over a third of the women who responded (%) were not given any contact details at all and % of women who responded stated that support needs go unmet. Most women who responded (%) had access to a clinical nurse specialist at some point during their cancer journey. Women reported a range of information and support needs, reflecting different values, preferences and circumstances. However certain types of information and support needs were more commonly reported than others. Women were most likely to report information and support needs connected with their treatment and its side effects and their disease and prognosis (Beesley et al., 00; Browall et al., 00; Steele and Fitch 00; Fitch and Steele, 00). Power et al., (00) reported that many patients expressed a desire not to find out all the information they could about their condition, and they purposefully avoided dealing with it whenever possible as a coping strategy. Recommendations Offer all women with newly diagnosed ovarian cancer information about their disease, including psychosocial and psychosexual issues, that: o is available at the time they want it o includes the amount of detail that they want and are able to deal with o is in a suitable format, including written information if possible. Ensure that information is available about: o the stage of the disease, treatment options and prognosis o how to manage the side effects of both the disease and its treatments in order to maximise well being o sexuality and sexual activity o fertility and hormone treatment o symptoms and signs of disease recurrence o genetics, including the chances of family members developing ovarian cancer o self-help strategies to optimise independence and coping o where to go for support, including support groups o how to deal with emotions such as sadness, depression, anxiety and a feeling of a lack of control over the outcome of the disease and treatment. Linking evidence to recommendations The GDG placed a high value on patient support but recognised there were continuing variation and gaps in service support and delivery. The GDG felt this variation led to unmet needs which need to be overcome. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

94 0 0 0 There was good quality evidence highlighting the need for the relevant information, tailored to the needs of the individual women, to be offered to women at the time that most suits their individual practical and psychological needs. The GDG noted that immediately after diagnosis, a woman s most pressing information needs related to treatment, its side effects, the disease and her prognosis. Other information including psychosocial and psychosexual issues, although important was not ranked as highly at this time. The GDG therefore felt it was important to make recommendations on both of these areas. This clinical question was not considered amenable to health economic evaluation as there was no comparative analysis. References Beesley V, Eakin E, Steginga S, Aitken J, Dunn J and Battistutta D. (00) Unmet needs of gynaecological cancer survivors: implications for developing community support services. Psycho-Oncology (): -00 Browall M, Carlsson M and Horvath GG. (00) Information needs of women with recently diagnosed ovarian cancer--a longitudinal study. Eur J Oncol Nursing (): 00- Department of Health () Improving outcomes in gynaecological cancers. Service guidance. Available from _00 Fitch M and Steele R (00). Identifying supportive care needs of women with ovarian cancer. Canadian Oncology Nursing Journal, 0, -. Jefferies H. (00) Ovarian cancer patients: are their informational and emotional needs being met? J Clin Nursing (): - National Institute for Clinical Excellence (00) Improving supportive and palliative care for adults with cancer. NICE cancer service guidance. London: National Institute for Clinical Excellence Power J, Brown L and Ritvo P. (00) A qualitative study examining psychosocial distress, coping, and social support across the stages and phases of epithelial ovarian cancer. Health Care for Women International (): - Steele R and Fitch MI. (00) Supportive care needs of women with gynecologic cancer. Cancer Nursing (): - Target Ovarian. (00) Mapping the experiences of those living or working with ovarian cancer in the UK. The Target Ovarian Cancer Pathfinder Study. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

95 Appendix A cost-utility analysis of diagnostic investigations in primary care for women with symptoms of ovarian cancer Introduction Around,00 new cases of ovarian cancer are diagnosed each year in the UK (CancerResearch UK, 00) with an overall five-year survival of about 0% in women diagnosed with early disease (stage I-II) and % in women with advanced disease (stage III-IV) (Hamilton et al., 00). For women presenting with symptoms in primary care, accurate diagnostic information at this stage enables timely referral which subsequently plays a vital role in the choice of treatment and achievable survival. This clinical question was highlighted as a priority for economic analysis because of the large number of patients with symptoms suggestive of ovarian cancer. In addition, there are significant differences in costs and health outcomes associated with the different diagnostic pathways, as well as the considerable economic burden of treating ovarian cancer. Objective To assess the cost-effectiveness of diagnostic strategies in primary care for women presenting with symptoms suggestive of ovarian cancer. Methods Economic evaluations of a diagnostic investigation require evidence on a number of issues, including disease prevalence and test accuracy. Furthermore, the accurate estimation of cost-effectiveness of one diagnostic strategy over another requires the consideration of downstream treatment effects, health-related preferences (utilities), healthcare resource use and unit costs. Therefore, the evaluation was undertaken by synthesizing evidence from a number of different sources using decision analytic techniques.. Study population The population considered within the analysis consisted of women presenting in primary care with symptoms consistent with suspected ovarian cancer.. Perspective This analysis was carried out from the perspective of the UKs National Health Service (NHS), in line with NICEs methodological recommendations. Health outcomes were expressed in terms of quality-adjusted life-years (QALYs).. Interventions The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

96 0 0 Given the large number of different diagnostic tests and potential combinations, a decision was made at the outset to limit the number of interventions to those that were listed by the Guideline Development Group (GDG) in the PICO tables for this clinical question. In all, seven core strategies were evaluated. To capture downstream consequences following the initial referral, the members of the GDG were asked to identify clinical pathways that were reflective of current UK clinical practice (Table A.). Table A. Summary of diagnostic strategies Strategy Primary care diagnostic investigation(s) Pelvic examination Ultrasound* Serum CA Ultrasound CT scan Pelvic examination and serum CA Ultrasound CT scan Ultrasound Serum CA CT scan Pelvic examination and ultrasound Serum CA CT scan Serum CA and ultrasound CT scan Pelvic examination, serum CA and ultrasound Secondary care diagnostic investigation(s) (following referral) Serum CA and ultrasound CT scan CT scan * Only done where pelvic examination did not detect a suspicious mass.. Structure of the model A decision tree (Figure A.) was constructed outlining the seven strategies of interest: three of the strategies included a single first test and the remaining four strategies were combination tests. The model was constructed using TreeAge Pro (00) software. A Markov process was embedded in the decision tree to model recurrence of the disease and survival based on the results of the diagnostic tests and the subsequent management of women presenting with symptom(s) of ovarian cancer. A hypothetical cohort of women presenting with symptom(s) of ovarian cancer in the primary care setting was considered for the analysis. In the base case, it was considered that the starting age of the patient population in the model was 0 years of age, while further analyses considered a starting age of 0 years. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

97 Figure A. Diagnostic strategies in primary care Decision tree for accuracy of staging procedures and related complications The square node at the beginning of the decision tree shows graphically the seven diagnostic strategies (see Table A.) that have been defined as relevant to the decision problem (Figure A.). Independent of which diagnostic strategy is undertaken; patients may or may not have a suspicious mass. This way of structuring the model allows information about the prevalence of a suspicious mass and accuracy of the diagnostic procedures as reported in the systematic reviews of the clinical evidence related to diagnostic investigation in primary care (in terms of their sensitivity and specificity values (Hunink and Glasziou, 00)) to be used. Patients in whom the results of primary care investigation did not identify a suspicious mass were assumed to be discharged, with the exception of those undergoing pelvic examination as their primary care test. Patients in whom malignancy has been suspected are referred to secondary care for further investigation. Patients who have undergone pelvic examination (strategy ) as part of their initial investigation in primary care are referred to secondary care if the test outcome identifies a suspicious mass. Patients in whom pelvic examination did not identify an abnormality undergo ultrasound in primary care. The result of the ultrasound is used to decide whether to refer the patient to secondary care. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

98 The pathway of diagnostic investigations in secondary care depends in part on the type of diagnostic test performed in primary care. The diagnostic pathway for each strategy following referral was outlined by the GDG. In order to maintain consistency within the guideline, imaging procedures reflect the current guideline recommendations. Pelvic examination Patients following strategy (see Table A.) as part of their investigation pathway and where the initial test (pelvic examination) identified a suspicious mass, are referred to secondary care and undergo combination serum CA plus ultrasound as the next diagnostic tests. At this stage, patients in whom a suspicious mass was not detected following investigation in secondary care (i.e. combination of serum CA plus ultrasound), undergo a repeat of the same test within a month and are either referred for a computerised tomography (CT) scan (to confirmed ovarian cancer) or are discharged. Patients in whom a suspicious mass was detected undergo further investigation (in secondary care) with a CT scan, which may confirm the presence and extent of suspected ovarian malignancy. Serum CA; pelvic examination plus serum CA; ultrasound; pelvic examination plus ultrasound In the case of strategies,, and (see Table ), those referred to secondary care with a suspicious mass either undergo ultrasound (strategies and ) or serum CA (strategies and ). If the result of the ultrasound further identifies a suspicious mass, the patient undergoes a CT scan to confirm the presence of ovarian malignancy. Similarly, patients in whom a suspicious mass was not detected following ultrasound or serum CA undergo a repeat of the same test within a month and are either referred to undergo a CT scan (to confirmed ovarian cancer) or are discharged. Serum CA plus ultrasound; pelvic examination plus CA plus ultrasound Lastly, patients following strategies and (see Table A.)) where a suspicious mass was detected, are referred to secondary care and undergo a CT scan to assess the extent of the ovarian cancer or an alternate diagnosis. To capture the downstream consequences of each diagnostic strategy, a clinical pathway was outlined encompassing treatment options following confirmation of ovarian malignancy. As such, it was agreed that following a CT scan, a proportion of patients with confirmed ovarian malignancy, will undergo either a surgical procedure, pathological investigation (biopsy) or will receive supportive care (where the patient is not fit for further treatment/investigation). For the purpose of this model it was agreed that following surgical and pathological procedures patients would be classified as either having disease confined to the ovaries (FIGO stage Ia Ic) or disease which is not confined to the ovaries (FIGO stages II-IV). Furthermore, patients in whom the CT scan did not confirm ovarian malignancy, undergo further investigation to differentiate the nature of the suspicious mass. It was agreed that for the purposes of this model two subgroups of patients withouth confirmed ovarian malignancy would be considered: patients with a benign gynaecological problem (for example a simple cyst) and patients with colorectal malignancy. Treatment options were defined for each subgroup of patients. A summary of the key structural assumptions are listed in Box A.. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

99 Box A. Key Structural Assumptions In primary care With the exception of those undergoing pelvic examination, patients in whom no malignancy was suspected from initial tests are discharged with no further follow up Patients who undergo pelvic examination in primary care and have no suspicious malignancy are re-tested using ultrasound In secondary care Patients in whom further investigation showed no suspicion of malignancy are re-tested within a month Computerised tomography scan is able to differentiate between ovarian and non-ovarian masses Histopathological tests are assumed to be 00% accurate.. Markov process to model prognosis of patients in the long term A Markov process was embedded in the decision tree to reflect the prognosis of patients according to the management received following the test results. In a Markov process a patients possible prognosis is divided into a series of discrete health states. Costs and benefits are assigned to each health state and transition probabilities are defined to model the movement of an individual between these health states over a particular time frame (cycle length). The costs and benefits of comparative treatments are then estimated on the basis of the length of time individuals spend in each health state. The aim of introducing a Markov process at the end of the decision tree was to reflect the pattern of recurrences and survival of patients in a simplified way, depending on whether the diagnostic investigation had been accurate in identifying a suspected mass and, consequently, whether patients were appropriately managed according to their true condition. Three heath states were considered for patients in whom malignancy is confined to the ovaries and who have completed treatment: remission, recurrence and death (all causes). For patients with advanced disease only two health states were considered: remaining in the advanced (recurrence) disease state or death. On each given cycle, patients with confined disease could remain in the disease-free state (remission), have a recurrence and progress to advanced disease or die. Patients with advanced disease could either remain in the advanced stage or die. Patients in whom colorectal malignancy was identified could either remain in that disease stage (Dukes stage A-D), progress or die. Two health states were considered for patients who have undergone treatment for a benign gynaecological problem, who require no further treatment or were discharged following a negative test outcome: patients could either remain alive or die. A one-year cycle length was used in all instances. The different probabilities of moving from one health state to another depend on the associated risk of recurrence, disease progression and death. Death can result from The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

100 0 0 0 ovarian cancer (if the patient had progressed), colorectal malignancy, or from all other causes.. Clinical evidence Economic modelling is a useful tool to synthesise data derived from multiple sources, given the fact that all the relevant costs and benefits of an intervention are rarely accurately captured by one single study. Although randomised controlled trials are usually the most reliable sources of evidence, they are not always available. Data is often used from non-randomised studies or from expert opinion in which case transparency and consistency is essential. Conducting a sensitivity analysis examines the robustness of the results obtained and the variables most likely to influence the results.. Data inputs.. Prevalence and test accuracy The clinical evidence required to populate the model was obtained from the systematic reviews conducted within the ovarian cancer guideline. The prevalence of the disease in primary care was assumed to be a linear summation of the prevalence of ovarian and colorectal malignancies and benign gynaecological problems. The estimates of prevalence of ovarian and colorectal malignancies are obtained from published literature (CancerResearch UK, 00; Hamilton et al., 00). GDG consensus was used to estimate the prevalence of benign gynaecological problems. The accuracy of the diagnostic procedures, in terms of the corresponding sensitivity and specificity values, was obtained from the systematic reviews of the clinical evidence conducted for this guideline (see clinical evidence in sections. and.). The accuracy of combination strategies were calculated assuming conditional independence. A summary of the estimates of disease prevalence and test accuracy used to populate the model are reported in Table A.. Table A. Disease prevalence and test accuracy Parameter description Parameter estimate Data source Disease Disease prevalence Data source Ovarian cancer 0.% Hamilton et al., 00 Benign gynaecological problem % Range (0% - 0%) GDG consensus Colorectal cancer 0.0% CancerResearchUK, 00 Test accuracy Sensitivity Specificity Data source Pelvic examination Myers et al., 00 Serum CA Myers et al., 00 Ultrasound Liu et al., 00 Combination tests Pelvic examination + CA Derived from single test estimates assuming test The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 00 of

101 Pelvic examination + ultrasound CA + ultrasound independence (see section. of the Evidence Review) Pelvic examination CA + ultrasound Secondary care test CT scan Liu et al., 00.. Proportion estimates The proportion of patients in each treatment arm, as defined by the model structure, was not consistently reported in the published literature. Therefore, proportions were estimated by the GDG. The estimates of the proportions are shown in Table A.. Table A. Estimates of proportions Parameter description Estimate (%) Patients in whom no cancer of the ovaries was detected following secondary care test : Proportion of patients who are diagnosed with a benign gynaecological problem (for example a simple cyst) Proportion of patient who are diagnosed with other cancer (colorectal) Patients in whom cancer of the ovaries was detected following secondary care test : Proportion of patients undergoing percutaneous biopsy (or any other histopathological investigation) Proportion of patients undergoing surgery 0 Proportion of patients who are not fit to undergo any further investigation and receive supportive care Patients who have undergone surgery : Proportion of patients in whom disease is confined to the ovaries (stage I) 0 Proportion of patients in whom disease is not confined to the ovaries (stage II-IV) 0 Patients with disease confined to the ovaries : Proportion of patients undergoing chemotherapy (carboplatin) 0 Proportion of patients who do not require further treatment (following surgery) and receive follow-up care Patients with disease not confined to the ovaries : 0 Proportion of patients undergoing chemotherapy (paclitaxel/carboplatin) Proportion of patients undergoing chemotherapy (paclitaxel/carboplatin) and further surgery 0 Estimation is based on an assumption that of all patients in whom cancer of the ovaries is detected: % will have advanced stage disease and % will have early stage disease (Kosary ; Bell et al., ). Of those with advanced stage disease 0% will undergo surgery and 0% biopsy. stage I includes stages Ia- Ic. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 0 of

102 Proportion of patients who are not fit for further treatment (following staging surgery) and are receiving supportive care Source: GDG Consensus; Warwick et al Treatment Surgery Historically, the mainstay of treatment for ovarian cancer was surgical excision. It has been estimated that the majority of patients with early and about half with advanced stage disease will require some form of surgery (Bell et al., ; Kosary ). For the purpose of this model, the GDG agreed that the majority of patients, in both groups, will undergo laparotomy with intent to perform total abdominal hysterectomy (TAH)/bilateral salpingo-oophorectomy (BSO)/omentectomy/peritoneal washings. In patients where no malignancy was suspected (for example, a simple cyst) it was agreed to assume the same procedures would be carried out. Mortality and morbidity rates associated with these surgical procedures were obtained from the published literature (Chien et al., 00; Gerestein et al., 00; Loft et al., ; Venesmaa and Ylikorkala ) or through GDG consensus and are shown Table A.. Table A. Mortality and morbidity associated with laparotomy Confined to the ovaries (stages a- c) Not confined to the ovaries (stages II-IV) Benign gynecological problem Mortality % % 0.% Morbidity %* 0-%* %** Source: Venesmaa et al., (); Gerestein et al., (00) (stage II-IV); Loft et al., () (benign problem); * GDG consensus; ** Chien et al., (00) Chemotherapy Within the guideline, a review of the clinical evidence was conducted to ascertain the most effective chemotherapy regimen in patients with early disease. To assure consistency between the guideline as a whole and the economic model, it was agreed that for the purposes of economic analysis, patients in whom cancer is confined to the ovaries receive a carboplatin-based chemotherapy regimen. Dosage, duration of treatment, estimates of overall survival and progression free survival were obtained from the ICON trial (Swart et al., 00)) (Table A.). The study did not report major toxicities associated with carboplatin. Patients with advanced disease (i.e. where cancer is not confined to the ovaries) followed the treatment pathway outlined by Guidance on the use of paclitaxel in the treatment of ovarian cancer (NICE, 00). Similarly, estimates of overall survival, progression free survival, duration of treatment and dosages of a combination of agents were taken from Bagnall et al., (00) (see Table A. below). Table A. Dosage, duration of treatment and survival estimates assumed by the model Confined to the ovaries Not confined to the ovaries Agent (s) Carboplatin Paclitaxel/carboplatin Dosage AUC mg/m AUC Number of cycles Progression free survival % (0 years PFS). months (median) (PFS) Overall survival (OS) % (0 years OS). months (median) The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 0 of

103 0 0 Data source ICON Trial (Swart et al., 00).. Supportive care and follow-up monitoring ICON Trial (Bagnall et al., 00) Supportive care No studies were found to provide estimates of healthcare resource use for the provision of supportive care specifically in this group of patients. Given the advanced stage of the disease, it was agreed that a patient will spend a third of their time at home, a third in a hospital and the latter stage in a hospice. For the purpose of this analysis, we obtained estimates of unit costs of resource use by GDG consensus. Follow-up monitoring Similarly, no studies were found quantifying healthcare resource use associated with the follow-up monitoring of women who had undergone treatment (surgery and chemotherapy). Other guidelines were used to identify relevant components of care and a likely schedule of follow-up monitoring for women who have undergone active treatment. The GDG agreed that follow-up monitoring should include a history and physical examination (including pelvic examination) every three months for three years and once a year for the following five years. Estimates of resource use were obtained by GDG consensus and are summarised in Table A.. Table A. Resource use associated with provision of supportive care and follow-up monitoring Supportive care (per patient) Number of units Hospital stay (in days) Hospice stay (in days) Home stay GP visits (0./week) District nurse Nurse specialist Follow-up monitoring (per year) Years - Physical examination (including pelvic examination) Years onwards Physical examination (including pelvic examination).. Other cancer colorectal It was agreed that for the purposes of this economic model estimates of survival associated with treatment for colorectal cancer will be used as proxy for the subgroup of patients in whom a non-gynaecological cancer was identified following diagnostic investigation. A summary of average survival (by stage) is reported in Table A.. Table A. Distribution and survival by stage (at diagnosis) Disease stage Proportion (NCIN, 00) Average Survival (Tappenden et al., 00) Dukes A.% years Dukes B.% years The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 0 of

104 0 0 0 Dukes C.%. years Dukes D.0%. years.. Health benefits The health benefits derived from using the alternative diagnostic strategies compared in the analysis were estimated in terms of the number of quality-adjusted life years (QALYs) gained. The base case analysis considered a lifetime horizon, although a shorter time horizon was considered in the sensitivity analysis. Markov processes were used to estimate life expectancy and QALYs gained by four different patient subgroups: Patients who were considered to have a suspicious mass at the beginning of the model (following initial test) and have undergone an appropriate treatment (true positive) Patients who did not have a suspicious mass at the beginning of the model (following initial test) but have undergone treatment after being wrongly diagnosed (false positive) Patients who did not have a suspicious mass at the start of the model (following initial test) and were discharged (true negative) Patients who have a suspicious mass at the start of the model (following initial test) but were wrongly discharged following diagnostic investigation (false negative). Estimates of life expectancy The transition probabilities of moving across health states (Figures A.-A.) were estimated from published studies (International Collaborative Ovarian Neoplasm Group, 00; Swart et al., 00), which reported rates of remission, recurrence and death following chemotherapy treatment in patients with localised and advanced disease. An appropriate adjustment was conducted to obtain yearly transition probabilities of recurrence and death in this subgroup of patients (Hunink and Glasziou, 00). Moreover, the transition probabilities were assumed to be constant throughout the time horizon of the model. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 0 of

105 Figure A. Markov process for prognosis of patient with early disease Figure A. Markov process for prognosis of patient with advance disease The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 0 of

106 Figure A. Markov process for prognosis of patient with colorectal cancer 0 For patients who did not have the disease, had a benign condition or required followup monitoring after undergoing chemotherapy, transition probabilities of moving from alive to dead from all causes were estimated using the age-related mortality rates (as reported by the Office of National Statistics, 00). For patients who are diagnosed with colorectal malignancy, progression from initial stage to the next or to death was captured by the transition probabilities reported in Tappenden et al., (00). The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 0 of

107 A summary of all transition probabilities used to populate the model is reported in Table A.. Table A. Transition probability between health states Transition probability Mean Description Ovarian cancer tprem_adv 0.0 Probability of recurrence (early disease) -tprem_adv 0. Probability of remaining in remission tpadv_death 0. Probability of remaining in the advanced disease state tpadv_death 0.0 Probability of dying (advanced disease) Colorectal cancer tpcrc_a_b 0. Probability of moving from Dukes A to Dukes B tpcrc_a_death 0 Probability of dying (Dukes A) tpcrc_b_c 0. Probability of moving from Dukes B to Dukes C tpcrc_b_death 0.0 Probability of dying (Dukes B) tpcrc_c_d 0. Probability of moving from Dukes C to Dukes D tpcrc_c_death 0.00 Probability of dying (Dukes C) tpcrc_d_death 0. Probability of dying (Dukes D) Utility estimates The value of estimating the number of QALYs gained is that this single measure combines the gains from mortality (quantity gains) and from morbidity (quality gains) (Drummond et al., 00). An index based on an individual s preference for a specific health state in relation to alternative health states (utility weights) were required in the model to estimate quality-adjusted life years (QALYs), which are calculated by weighting life expectancy by a measure of associated health-related quality of life. Estimates of health state utilities specific to ovarian cancer patients were obtained from published studies. There are a number of studies that report utility weights associated with diagnostic investigations and treatments of ovarian cancer. Havrilesky et al., (00) reported utility estimates related to various heath states following false positive/negative test results and treatment with toxicities. Utility estimates obtained using the time trade-off method (TTO) tended to be slightly higher compared to those obtained using a visual analogue score (VAS). Drummond et al., (00) noted that visual scales for comparing health state preferences are subject to inherent biases and are generally less accurate. For this reason we used utility estimates derived using the TTO method. Utility estimates associated with undergoing surgery and colorectal cancers were obtained from Grann et al., () and Tappenden, et al. (00) respectively. The utility values used in the model are summarised in Table A. below. Table A. Utility values Health state Mean Data Source Diagnostic test false positive/negative result 0. Havrilesky et al., 00 The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 0 of

108 0 0 Chemotherapy (carboplatin) 0. Havrilesky et al., 00 Chemotherapy (paclitaxel) 0. Havrilesky et al., 00 Toxicity grade - (paclitaxel) 0. Havrilesky et al., 00 Surgery 0. Grann et al., Recurrence 0. Havrilesky et al., 00 Remission (early) 0. Havrilesky et al., 00 Stable - advanced disease 0. Grann et al., Colorectal cancer (by stage) Dukes A 0. Tappenden et al., 00 Dukes B 0.0 Tappenden et al., 00 Dukes C 0.0 Tappenden et al., 00 Dukes D 0. Tappenden et al., 00 Supportive care 0. Havrilesky et al., 00 Follow-up 0. Assumed.. Cost estimates The costs considered in this analysis were only those relevant to the UK NHS, in accordance with the perspective taken by the NICE Reference Case for economic evaluations. Costs were estimated based on 00- prices. When costs have been taken from other sources and are applicable to a different price year, they have been inflated using the Hospital and Community Health Services Pay and Prices Index (PSSRU, 00). The categories of costs included: Cost of diagnostic tests (in primary and secondary care) Cost of therapy (surgery, drug acquisition costs, administration costs) Cost of major treatment related to morbidity Cost of healthcare resource use associated with supportive care and followup monitoring Costs of diagnostic tests The cost estimates of diagnostic tests relevant to this analysis were obtained from various sources. Unit costs of ultrasound, CT and MRI were obtained from the NHS Reference Costs and estimated at, and respectively (HRG codes: RAZ, RAZ and RA0Z). The cost of pelvic examination was estimated using unit cost reported in the Personal Social Services Research Unit (PSSRU 00) and included the cost of GP s and nurses time. Unit costs of tumour marker test (serum CA) was estimated at and obtained using GDG consensus. Unit costs of combination tests were estimated as a sum of the unit costs of the individual tests. The cost estimates of pathological investigation were assumed to consist of the cost of percutaneous biopsy and aspiration cytology. These costs were obtained from The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 0 of

109 0 0 NHS Reference costs and from GDG consensus, and were estimated to be, and respectively. A summary of unit costs of diagnostic tests are presented in Table A.0. Table A.0 Cost estimates of diagnostic tests Mean ( ) Data source Ultrasound NHS Reference Cost: HRG code RAZ Pelvic examination GP practitioner (per procedure) PSSRU 00 GP nurse (per procedure) 0 PSSRU 00 Total Serum CA GDG consensus Cost estimation of combination diagnostic tests Pelvic examination + ultrasound Pelvic examination + serum CA Serum CA + ultrasound Pelvic examination + ultrasound + serum CA CT scan NHS Reference Cost: HRG code RAZ Biopsy Percutaneous biopsy NHS Reference Cost: HRG code FZC Aspiration cytology GDG consensus Total MRI NHS reference Cost: HRG code RA0Z Cost of Treatment Chemotherapy The drug costs were calculated for chemotherapy regimens for patients with localised and advanced disease, assuming that a patient received one dose per -week cycle for single or combination therapy (Table A.). In addition to the drug acquisition costs, the cost of administering the drug was estimated from the NHS Reference Costs. Administration of carboplatin and the carboplatin/paclitaxel combination regimens was assumed to be performed on an outpatient basis. The cost of administering these regimens was estimated using outpatient tariffs of (HRG SBZ) and (HRG SBZ) respectively. This cost includes hospital overheads, the administration costs of chemotherapy and clinical time. These assumptions were verified with the GDG. The base case analysis used list prices for drugs obtained from the British National Formulary (BMG Group and Pharmaceutical Press, 00). The effect of the drug discounts were explored through sensitivity analysis. Table A. Drug acquisition costs Strategy Carboplatin Carboplatin/paclitaxel Carboplatin Carboplatin Paclitaxel List prices, (BNF, 00) The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 0 of

110 ml vial. 0 ml vial.. 0 ml vial ml vial 0 0 i.v. concentrate (mg/ml) 0 0 Recommended dose (mg/m ) 0 Average cost per vial ( )... Number of vials Average drug cost per cycle ( )... Surgery Patients identified as having ovarian cancer or a benign gynaecological problem undergo a surgical procedure. The unit costs considered in this analysis were estimated by mapping the Classification of Surgical Operations and Procedures from the Office of Population, Censuses and Survey (OPCS ) into Health Related Groups (HRGs) and by identifying the relevant unit cost as reported in the NHS Reference Costs for the specific HRGs. OPCS codes for laparotomy for malignant and benign conditions were obtained via GDG consensus. Costs of surgical procedures for malignant and benign gynaecological problems are reported in Table A.. Table A. Costs of surgical procedures Mean ( ) Data source Laparotomy with malignancy (no complications), NHS Reference Cost: HRG code MA0Z Laparotomy with malignancy (with complications),0 NHS Reference Cost: HRG code MA0Z* Laparotomy without malignancy (no complications), NHS Reference Cost: HRG code MA0B Laparotomy without malignancy (with complications),0 NHS Reference Cost: HRG code MA0A 0 * Extra cost associated with complication was obtained using percentage change between HRG MA0A and MA0B as a proxy. Treatment of colorectal cancer Lifetime costs estimates of the treatment of colorectal cancer were obtained from a published study by Tappenden et al., (00) and are reported in the Table A. below. Table A. Lifetime costs of treatment of colorectal cancer Disease stage Mean cost ( ) Dukes A, Dukes B, Dukes C,0 Dukes D, Source: Tappenden et al., 00 The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page 0 of

111 0 Cost of supportive care and follow-up monitoring No published data was found that quantified healthcare resource use associated with the provision of supportive care and follow up monitoring specifically in patient subgroups identified in the model. Categories and number of units of relevant resource use items were obtained via GDG consensus. The total number of units for each category of resource use was multiplied by the cost of providing it (PSSRU, 00). A summary of unit costs for each category of resource use are shown in Table A.. Table A. Unit cost of supportive care resource use Resource Unit cost ( ) Data source Hospital specialist palliative care NHS Reference costs: HRG code SD0A support Hospice specialist palliative care NHS Reference costs: HRG code SD0A GP visits PSSRU, 00 District nurse PSSRU, 00 Nurse specialist PSSRU, 00 Annual follow-up monitoring Years - PSSRU, 00 Year (onwards) PSSRU, Discounting Within health economic evaluation, the discounting of costs and health outcomes is standard practice since costs and benefits that accrue in the future are given less weight to those which occur in the present. Following NICE methodological guidance (NICE, 00), all costs and health outcomes are discounted at.% per year.. Sensitivity analysis A series of one-way sensitivity analyses were conducted to assess the robustness of the study results. One-way sensitivity analysis describes the process of changing one parameter in the model and re-running the model to see how a change in this parameter influences overall results. Five scenarios were considered and are detailed below: Nationally-agreed drug discounts in England were as follows: the cost per dose of paclitaxel is. compared to a list price of per dose (NHS Purchasing and Supplies Agency, August 00). The price of carboplatin is. compared to a list price of per dose. In Wales, nationally-agreed discounts were: % per dose for paclitaxel and % for carboplatin (personal communication from the Welsh Health Supplies, August 00). Based on these rates, the discounted cost of each regimen was calculated for England and for Wales. Whilst it is acknowledged that regional pharmacies and/or commissioners may negotiate other discounts separately, only The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

112 0 nationally agreed discounts are considered (NICE, 00). The average discounted cost across both regions is also reported in Table A.. The prevalence of ovarian malignancy in primary care was decreased to 0.%. The prevalence of benign gynaecological problem was varied over an agreed range (0% - 0%). The proportion of patients who are not fit for further treatment following diagnostic investigation was decreased to %. The age at the start of the model was increased from 0 to 0 years of age. Table A. Discounted drug acquisition costs in England and Wales Regimen Carboplatin Carboplatin/paclitaxel Average cost of regimen per cycle ( ) List price.. Discount price (England) Discount price (Wales) 0 0 However these scenarios are unlikely to happen independently; they are more likely to occur concurrently. To fully characterise this uncertainty and to estimate the effects of the parameter uncertainty on the results, a probabilistic sensitivity analysis (PSA) was undertaken. Firstly, the stochastic parameters in the model were identified (presented in the first column of Table A.). These are parameters which are (arguably) measureable, but are associated with sampling uncertainty. Secondly, these parameters were specified as distributions rather than point estimates (see fourth column of Table A.). Distributions associated with each of these parameters were selected according to a well developed body of methodological literature. The data required to inform these distributions was taken from the same sources as was used for the point estimates. Parameters not chosen for PSA: unit costs of health professionals and drug acquisition estimates of test accuracy Thirdly, the analysis was run 0,000 times. For each simulation, different values were picked from the various distributions for each stochastic parameter in the model. Table A. Parameters varied in the probabilistic sensitivity analysis Parameter Utilities Deterministic value Distribution assigned Source Diagnostic test false positive/negative result 0. Havrilesky. et al., 00 The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

113 Stable advanced disease 0. Beta Grann et al., Advanced (undergoing chemotherapy) 0. Beta Havrilesky. et al., 00 Advanced (undergoing chemotherapy with toxicity) 0. Beta Havrilesky. et al., 00 Early (chemotherapy) 0. Beta Havrilesky. et al., 00 Early (recurrence) 0. Beta Havrilesky. et al., 00) Early (remission) 0. Beta Havrilesky. et al., 00 Surgery 0. Beta Grann et al., Colorectal cancer Dukes A 0. Beta Tappenden et al., 00 Colorectal cancer Dukes B 0.0 Beta Tappenden et al., 00 Colorectal cancer Dukes C 0.0 Beta Tappenden et al., 00 Colorectal cancer Dukes D 0. Beta Tappenden et al., 00 Supportive care 0. Beta Havrilesky. et al., 00 Follow-up 0. Beta Assumed Transition probability tpadv_dead 0.0 Beta Bagnall et al., 00 tprem_recadv 0. Beta Swart et al., 00 tpcrc_a_b 0. Dirichlet Tappenden et al., 00 tpcrc_a_death 0 Dirichlet Tappenden et al., 00 tpcrc_b_c 0. Dirichlet Tappenden et al., 00 tpcrc_b_death 0.0 Dirichlet Tappenden et al., 00 tpcrc_c_d 0. Dirichlet Tappenden et al., 00 tpcrc_c_death 0.0 Dirichlet Tappenden et al., 00 tpcrc_d_death 0. Dirichlet Tappenden et al., 00 Proportions and rates Prior disease prevalence 0. Beta Hamilton et al., 00 Rate of toxicity (alopecia in advanced stage) 0. Beta Bagnall et al., 00 Rate of mortality (early) post surgery 0.0 Beta Venesmaa et al. Rate of mortality (advanced) - post surgery 0.0 Beta Gerestein et al., 00 Rate of mortality (benign) post surgery 0.00 Beta Loft et al., The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

114 Rate of morbidity (early) post surgery 0.0 Beta GDG consensus Rate of morbidity (advanced) - post surgery 0. Beta GDG consensus Rate of morbidity (benign) post surgery 0.0 Beta Chien et al., Proportion of patients with disease confined to the ovaries (undergoing treatment) Proportion of patients in whom ovarian cancer is detected (following secondary care test) Proportion of patients with disease not confined to the ovaries (undergoing treatment) Proportion of patients with benign gynaecological problem 0. Beta GDG consensus (0.; 0.0; 0.0) (0.; 0.; 0.0) Dirichlet Dirichlet GDG consensus GDG consensus 0. Beta GDG consensus Proportion of patients with colorectal cancer 0. Beta GDG consensus 0 Proportion of Dukes A-D (0.; 0.; 0.; 0.) Results Dirichlet Tappenden et al., 00 A summary of expected costs and effects associated with each diagnostic strategy in the model are presented in Table A.. The expected cost of the strategies varies widely, ranging from the least expensive (serum CA) at just over,00 to the most expensive (combination strategy of pelvic examination plus serum CA plus ultrasound) at,0 per patient. Health outcomes, measured in terms of QALYs, ranged from 0. for the serum CA strategy to. for pelvic examination plus serum CA plus ultrasound combination strategy. Serum CA (single test) strategy on average generates 0. QALYs and ultrasound (single test) generates 0. a difference of 0.00 QALYs is an equivalent (on average) of an additional. days of perfect health. Table A. Base case total expected cost and QALYs Strategy Cost ( ) Effectiveness (QALY) Serum CA,. 0. ICER Ultrasound,0. 0. (Dominated) Pelvic examination + serum CA, (Dominated) Pelvic examination + ultrasound,. 0. (Dominated) Pelvic examination,. 0. (Dominated) Serum CA + ultrasound,0.. (Dominated) Pelvic examination + ultrasound + serum CA,0.. (Dominated) ICER incremental cost-effectiveness ratio All strategies in this analysis are dominated by the serum CA strategy. A strategy is said to be dominated if it is both more costly and less effective than its comparator. Graphical representation of the base case shown on Figure A.. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

115 Figure A. Cost-effectiveness plane for base-case results 0. Sensitivity analysis The results of base case analysis were not sensitive to any of the five scenarios outlined above in section.. The discount on paclitaxel and carboplatin available in England and Wales is considerable; the price is about 0% of the list price. This drastically reduced the costs attributed to marginal reduction in the overall expected costs for each of the strategies, but did not alter the ranking of the cost-effective diagnostic strategies (Table A.). Table A. One-way sensitivity analysis drug discounts Strategy Costs ( ) Costs ( ) Effectiveness (QALY) England Wales Serum CA,.,. 0.0 Ultrasound,.,. 0. Pelvic examination + serum,00., CA Pelvic examination + ultrasound,.,. 0. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

116 Pelvic examination,0.,0. 0. Serum CA + ultrasound,.,0..0 Pelvic examination + ultrasound + serum CA,.,.0. Similarly, the results of the one-way sensitivity analysis of the other scenarios (for example, changes in the prevalence, proportion of patients undergoing supportive care and starting age of the patients in the model) showed changes in the overall expected costs and health benefits but did not alter the ranking of the cost-effective diagnostic strategy. The results of deterministic sensitivity analysis are presented in Tables A. and A.0. Table A. One-way sensitivity analysis change in prevalence Strategy Prevalence of ovarian cancer 0.% Prevalence of benign condition 0% Prevalence of benign condition 0% Costs ( ) QALYs Costs ( ) QALYs Costs ( ) QALYs Serum CA,. 0.0,. 0., Ultrasound,. 0.,. 0.,. 0. Pelvic examination +,0. 0.,. 0.,. 0.0 serum CA Pelvic examination +,. 0.0,. 0., ultrasound Pelvic examination,0. 0.,. 0.0, Serum CA + ultrasound Pelvic examination + ultrasound + serum CA,..,0..,..,..,0..,.. Table A.0 One-way sensitivity analysis proportion estimates and starting age Strategy Prop. Supportive Care % Starting age 0 years Costs ( ) QALYs Costs ( ) QALYs Serum CA,. 0.0,..0 Ultrasound,0. 0.,0..0 Pelvic examination + serum CA,0. 0.,0.0.0 Pelvic examination + ultrasound,. 0.,.. Pelvic examination,. 0.,.. Serum CA + ultrasound,.0.0,.. Pelvic examination + ultrasound + serum CA,..,0.0. The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of

117 0 To fully assess the effects of the parameter uncertainty on the results, the base case model was estimated using probabilistic sensitivity analysis. As with the deterministic results, the results of PSA showed serum CA as the dominant strategy. The corresponding cost-effectiveness acceptability curve (CEAC) shows that, at a threshold of 0,000 per QALY, the probability that the serum CA strategy is the most cost effective option is almost %. Moreover, the serum CA strategy had the highest probability of being the most cost-effective when compared to other strategies, at any level of willingness-to-pay per additional QALY gained (Figure A.). Figure A. Cost-effectiveness acceptability curve for base case results 0 0 PE = pelvic examination; CA = serum CA; USS = ultrasound Discussion The aim of this study was to assess the cost-effectiveness of diagnostic strategies for women presenting with symptoms suggestive of ovarian cancer in primary care. A cost-utility analysis was undertaken to estimate the incremental cost per QALY of seven diagnostic strategies, which included the downstream costs and consequences of subsequent treatments considered likely to reflect current UK clinical practice and to be consistent with recommendations made within this guideline. The base-case results of this analysis provide a clear message for recommendations on this topic, in terms of cost-effectiveness. They show that the serum CA diagnostic strategy dominates all other strategies. The robustness of the model was tested using one-way sensitivity analysis. The results of the deterministic sensitivity analysis showed that although expected costs and health outcomes varied across strategies, the overall ranking of the cost-effective strategy did not change. Moreover, probabilistic sensitivity analysis was undertaken to fully assess the effects of the parameter uncertainty on the results. The results of the PSA showed serum CA The recognition and initial management of ovarian cancer: full guideline DRAFT (September 00) Page of